Appendix D — Further explanations

For each main section of the book, open-access reviews, tutorials and textbook chapters that explain the topic in more depth than this guide. They were selected for clarity, not for completeness, and all can be read without a subscription. Each entry gives the title, a DOI or web link, the licence and, where available, the abstract. In the PDF version, one or two particularly useful papers per chapter are reproduced in full at the end of the chapter’s section, under their original licences.

D.1 Sequence searching

Further explanations for Chapter 1.

D.1.1 Homology and sequence similarity

• Glover et al. (2019) — A community review of how orthologues and paralogues are defined, inferred and benchmarked. Useful background for the distinction between homology, orthology and paralogy made in this section. (review; CC BY 4.0)

  Advances and Applications in the Quest for Orthologs. Glover et al. (2019). Molecular Biology and Evolution. doi:10.1093/molbev/msz150

  Abstract

  Gene families evolve by the processes of speciation (creating orthologs), gene duplication (paralogs), and horizontal gene transfer (xenologs), in addition to sequence divergence and gene loss. Orthologs in particular play an essential role in comparative genomics and phylogenomic analyses. With the continued sequencing of organisms across the tree of life, the data are available to reconstruct the unique evolutionary histories of tens of thousands of gene families. Accurate reconstruction of these histories, however, is a challenging computational problem, and the focus of the Quest for Orthologs Consortium. We review the recent advances and outstanding challenges in this field, as revealed at a symposium and meeting held at the University of Southern California in 2017. Key advances have been made both at the level of orthology algorithm development and with respect to coordination across the community of algorithm developers and orthology end-users. Applications spanned a broad range, including gene function prediction, phylostratigraphy, genome evolution, and phylogenomics. The meetings highlighted the increasing use of meta-analyses integrating results from multiple different algorithms, and discussed ongoing challenges in orthology inference as well as the next steps toward improvement and integration of orthology resources.

D.1.2 Dynamic programming and sequence alignment

• Kellis et al. (2020d) — Lecture-note style chapter (based on MIT course 6.047) that formulates the alignment problem and derives the dynamic-programming solution and the Needleman–Wunsch algorithm step by step, with a short look at multiple alignment. (book chapter; CC BY-NC-SA 4.0)

  Sequence Alignment and Dynamic Programming (Chapter 2 of Computational Biology: Genomes, Networks, and Evolution). Kellis et al. (2020). LibreTexts (MIT open textbook). https://bio.libretexts.org/Bookshelves/Computational_Biology/Book%3A_Computational_Biology_-_Genomes_Networks_and_Evolution_(Kellis_et_al.)/02%3A_Sequence_Alignment_and_Dynamic_Programming

D.1.3 Heuristic search methods

• Kerfeld and Scott (2011) — A short educational article that explains BLAST scores and E-values and how they are used to argue for evolutionary relationships. A gentle introduction to the alignment statistics discussed in this section. (tutorial; CC BY 4.0)

  Using BLAST to Teach “E-value-tionary” Concepts. Kerfeld & Scott (2011). PLoS Biology. doi:10.1371/journal.pbio.1001014

  Abstract

  BLAST (Basic Local Alignment Search Tool; [1],[2]) is the key bioinformatic tool for sequence comparison and retrieval from databases. BLAST is often the first step in using sequence-based information to design experiments and contextualize experimental results. Its speed and ease of use help account for this, with experiments requiring simply submitting a sequence of interest (the query sequence) and waiting a few seconds. At the same time, by rendering thinking unnecessary, BLAST is often used suboptimally, with many experienced researchers simply using the default parameters because they do not know how to manipulate them or accepting results with little understanding of their full meaning (or lack thereof). Recognition of the importance of BLAST to modern life sciences has led to its use in many biology courses, even at the high school level, to introduce students to bioinformatics applications in the life sciences. Concepts of molecular evolution (e.g., gene duplication and divergence; orthologs versus paralogs) are quite abstract and are best communicated with examples (Box 1). It is possible to use case studies from the literature, but the abundance of sequence data present in public databases raises the far more attractive possibility of using searches tailored to a particular course, or, better yet, allowing the students to choose their own examples. Box 1. Concepts at a Glance Leads into biological sciences Molecular evolution (e.g., homologs, paralogs, orthologs) Gene alignment Domain structure of proteins Biochemical nature of amino acids: “frequent” and “infrequent” substitutions Conserved versus divergent regions of genes Leads into math Amino acid identity matrices Quantification of sequence identity and similarity Leads into information technology Database queries and their tradeoffs (speed versus completeness) Large dataset management Less obviously, another benefit of teaching students how the BLAST algorithm works is that it provides an opportunity to illustrate how mathematics functions as a language of biology. For example, given that BLAST has been designed to retrieve homologs, there are several steps in the BLAST program that incorporate molecular evolution concepts to maximize the possibility of finding sequences with a shared evolutionary history. More basically, understanding the steps in the calculation of an E-value provides an opportunity to show the relationship between how the algorithm works and fundamental principles of biochemistry and evolution. Here, we provide an approach to teaching the basics of BLAST to students in order to emphasize how the algorithm translates fundamental biological principles into numerical terms culminating in an E value. Acquiring a feel for the algorithm and exploring genomic sequence data has the potential to inform a student's grasp of biomedical, biochemical, and biogeochemical concepts, presenting an excellent opportunity for multidisciplinary integration.

• Kellis et al. (2020b) — Covers global, local and semi-global alignment, linear-time exact string matching and its preprocessing, the BLAST algorithm and the probabilistic foundations of alignment scores. (book chapter; CC BY-NC-SA 4.0)

  Rapid Sequence Alignment and Database Search (Chapter 3 of Computational Biology: Genomes, Networks, and Evolution). Kellis et al. (2020). LibreTexts (MIT open textbook). https://bio.libretexts.org/Bookshelves/Computational_Biology/Book%3A_Computational_Biology_-_Genomes_Networks_and_Evolution_(Kellis_et_al.)/03%3A_Rapid_Sequence_Alignment_and_Database_Search

• Alser et al. (2021) — A broad review of read-mapping algorithms, including hash tables, suffix arrays and the FM-index/Burrows–Wheeler transform. Useful for the exact-matching part of this section. (review; CC BY 4.0)

  Technology dictates algorithms: recent developments in read alignment. Alser et al. (2021). Genome Biology. doi:10.1186/s13059-021-02443-7

  Abstract

  Aligning sequencing reads onto a reference is an essential step of the majority of genomic analysis pipelines. Computational algorithms for read alignment have evolved in accordance with technological advances, leading to today's diverse array of alignment methods. We provide a systematic survey of algorithmic foundations and methodologies across 107 alignment methods, for both short and long reads. We provide a rigorous experimental evaluation of 11 read aligners to demonstrate the effect of these underlying algorithms on speed and efficiency of read alignment. We discuss how general alignment algorithms have been tailored to the specific needs of various domains in biology.

D.1.4 Alignment-free sequence comparison

• Zielezinski et al. (2019) — A community benchmark of many alignment-free methods (k-mer based, MinHash and others) on protein, gene-tree, genome and horizontal-transfer tasks. Shows where these methods work well and where they do not. (review; CC BY 4.0)

  Benchmarking of alignment-free sequence comparison methods. Zielezinski et al. (2019). Genome Biology. doi:10.1186/s13059-019-1755-7

  Abstract

  Background Alignment-free (AF) sequence comparison is attracting persistent interest driven by data-intensive applications. Hence, many AF procedures have been proposed in recent years, but a lack of a clearly defined benchmarking consensus hampers their performance assessment. Results Here, we present a community resource (http://afproject.org) to establish standards for comparing alignment-free approaches across different areas of sequence-based research. We characterize 74 AF methods available in 24 software tools for five research applications, namely, protein sequence classification, gene tree inference, regulatory element detection, genome-based phylogenetic inference, and reconstruction of species trees under horizontal gene transfer and recombination events. Conclusion The interactive web service allows researchers to explore the performance of alignment-free tools relevant to their data types and analytical goals. It also allows method developers to assess their own algorithms and compare them with current state-of-the-art tools, accelerating the development of new, more accurate AF solutions.

D.1.5 Profiles and multiple sequence alignments

• Notredame (2007) — A concise review of progressive, consistency-based and iterative multiple alignment methods and of how alignment accuracy is benchmarked. It covers the main ideas behind Clustal, T-Coffee, MUSCLE and related tools, though not the most recent methods. (review; CC BY 4.0)

  Recent Evolutions of Multiple Sequence Alignment Algorithms. Notredame (2007). PLoS Computational Biology. doi:10.1371/journal.pcbi.0030123

  Abstract

  An ever-increasing number of biological modeling methods depend on the assembly of an accurate multiple sequence alignment (MSA). These include phylogenetic trees, profiles, and structure prediction. Assembling a suitable MSA is not, however, a trivial task, and none of the existing methods have yet managed to deliver biologically perfect MSAs. Many of the algorithms published these last years have been extensively described [1–3], and this review focuses only on the latest developments, including meta-methods and template-based alignment techniques.

• Yoon (2009) — A tutorial-style review of HMMs, the Viterbi, forward and backward algorithms, parameter estimation and profile HMMs for sequence families. Fits the HMM and profile-HMM parts of this section. (tutorial; CC BY (Europe PMC))

  Hidden Markov Models and their Applications in Biological Sequence Analysis. Yoon (2009). Current Genomics. doi:10.2174/138920209789177575

  Abstract

  Hidden Markov models (HMMs) have been extensively used in biological sequence analysis. In this paper, we give a tutorial review of HMMs and their applications in a variety of problems in molecular biology. We especially focus on three types of HMMs: the profile-HMMs, pair-HMMs, and context-sensitive HMMs. We show how these HMMs can be used to solve various sequence analysis problems, such as pairwise and multiple sequence alignments, gene annotation, classification, similarity search, and many others.

• Kellis et al. (2020a) — Introduces Markov chains and HMMs with a casino example and describes the main HMM algorithms, close to the worked example used in this section. (book chapter; CC BY-NC-SA 4.0)

  Hidden Markov Models I (Chapter 7 of Computational Biology: Genomes, Networks, and Evolution). Kellis et al. (2020). LibreTexts (MIT open textbook). https://bio.libretexts.org/Bookshelves/Computational_Biology/Book%3A_Computational_Biology_-_Genomes_Networks_and_Evolution_(Kellis_et_al.)/07%3A_Hidden_Markov_Models_I

D.1.6 RNA homology search

• Kellis et al. (2020c) — Explains RNA secondary structure, dynamic-programming approaches to the folding problem, the evolution of RNA structure and a probabilistic approach to folding. (book chapter; CC BY-NC-SA 4.0)

  RNA Folding (Chapter 10 of Computational Biology: Genomes, Networks, and Evolution). Kellis et al. (2020). LibreTexts (MIT open textbook). https://bio.libretexts.org/Bookshelves/Computational_Biology/Book%3A_Computational_Biology_-_Genomes_Networks_and_Evolution_(Kellis_et_al.)/10%3A_RNA_Folding

• Zhao et al. (2021) — Reviews thermodynamic, comparative and machine-learning approaches to RNA secondary structure prediction. Background for the structure-prediction part of this section rather than for homology search itself. (review; CC BY 4.0)

  Review of machine learning methods for RNA secondary structure prediction. Zhao et al. (2021). PLOS Computational Biology. doi:10.1371/journal.pcbi.1009291

  Abstract

  Secondary structure plays an important role in determining the function of noncoding RNAs. Hence, identifying RNA secondary structures is of great value to research. Computational prediction is a mainstream approach for predicting RNA secondary structure. Unfortunately, even though new methods have been proposed over the past 40 years, the performance of computational prediction methods has stagnated in the last decade. Recently, with the increasing availability of RNA structure data, new methods based on machine learning (ML) technologies, especially deep learning, have alleviated the issue. In this review, we provide a comprehensive overview of RNA secondary structure prediction methods based on ML technologies and a tabularized summary of the most important methods in this field. The current pending challenges in the field of RNA secondary structure prediction and future trends are also discussed.

D.1.7 Databases and domains

• Das et al. (2015) — A short review, based on CATH, of how domain superfamilies diversify in structure and function. Useful for the structural classification of domains. (review; CC BY 4.0)

  Diversity in protein domain superfamilies. Das et al. (2015). Current Opinion in Genetics & Development. doi:10.1016/j.gde.2015.09.005

  Abstract

  Whilst ∼93% of domain superfamilies appear to be relatively structurally and functionally conserved based on the available data from the CATH-Gene3D domain classification resource, the remainder are much more diverse. In this review, we consider how domains in some of the most ubiquitous and promiscuous superfamilies have evolved, in particular the plasticity in their functional sites and surfaces which expands the repertoire of molecules they interact with and actions performed on them. To what extent can we identify a core function for these superfamilies which would allow us to develop a 'domain grammar of function' whereby a protein's biological role can be proposed from its constituent domains? Clearly the first step is to understand the extent to which these components vary and how changes in their molecular make-up modifies function.

• Forslund et al. (2019) — Explains what domains and domain architectures are, how they are annotated with resources such as Pfam and InterPro, and how architectures evolve by fusion, fission and shuffling. (book chapter; CC BY 4.0)

  Evolution of Protein Domain Architectures. Forslund et al. (2019). Evolutionary Genomics. doi:10.1007/978-1-4939-9074-0_15

  Abstract

  This chapter reviews current research on how protein domain architectures evolve. We begin by summarizing work on the phylogenetic distribution of proteins, as this will directly impact which domain architectures can be formed in different species. Studies relating domain family size to occurrence have shown that they generally follow power law distributions, both within genomes and larger evolutionary groups. These findings were subsequently extended to multi-domain architectures. Genome evolution models that have been suggested to explain the shape of these distributions are reviewed, as well as evidence for selective pressure to expand certain domain families more than others. Each domain has an intrinsic combinatorial propensity, and the effects of this have been studied using measures of domain versatility or promiscuity. Next, we study the principles of protein domain architecture evolution and how these have been inferred from distributions of extant domain arrangements. Following this, we review inferences of ancestral domain architecture and the conclusions concerning domain architecture evolution mechanisms that can be drawn from these. Finally, we examine whether all known cases of a given domain architecture can be assumed to have a single common origin (monophyly) or have evolved convergently (polyphyly). We end by a discussion of some available tools for computational analysis or exploitation of protein domain architectures and their evolution.

• Wang et al. (2021) — An overview of sequence- and structure-based methods and databases for identifying protein domains. Link only because of the ND licence. (review; CC BY-NC-ND 4.0 (link only))

  Protein domain identification methods and online resources. Wang et al. (2021). Computational and Structural Biotechnology Journal. doi:10.1016/j.csbj.2021.01.041

  Abstract

  Protein domains are the basic units of proteins that can fold, function, and evolve independently. Knowledge of protein domains is critical for protein classification, understanding their biological functions, annotating their evolutionary mechanisms and protein design. Thus, over the past two decades, a number of protein domain identification approaches have been developed, and a variety of protein domain databases have also been constructed. This review divides protein domain prediction methods into two categories, namely sequence-based and structure-based. These methods are introduced in detail, and their advantages and limitations are compared. Furthermore, this review also provides a comprehensive overview of popular online protein domain sequence and structure databases. Finally, we discuss potential improvements of these prediction methods.

D.1.8 Protein language models

• Chandra et al. (2023) — Explains the transformer architecture and its use in protein language models for predicting protein properties. A readable introduction for students without a machine-learning background. (review; CC BY 4.0)

  Transformer-based deep learning for predicting protein properties in the life sciences. Chandra et al. (2023). eLife. doi:10.7554/elife.82819

  Abstract

  Recent developments in deep learning, coupled with an increasing number of sequenced proteins, have led to a breakthrough in life science applications, in particular in protein property prediction. There is hope that deep learning can close the gap between the number of sequenced proteins and proteins with known properties based on lab experiments. Language models from the field of natural language processing have gained popularity for protein property predictions and have led to a new computational revolution in biology, where old prediction results are being improved regularly. Such models can learn useful multipurpose representations of proteins from large open repositories of protein sequences and can be used, for instance, to predict protein properties. The field of natural language processing is growing quickly because of developments in a class of models based on a particular model-the Transformer model. We review recent developments and the use of large-scale Transformer models in applications for predicting protein characteristics and how such models can be used to predict, for example, post-translational modifications. We review shortcomings of other deep learning models and explain how the Transformer models have quickly proven to be a very promising way to unravel information hidden in the sequences of amino acids.

• Valentini et al. (2023) — Surveys protein language models, their training and their uses in structure and function prediction and protein design. Gives broader context for this section. (review; CC BY 4.0)

  The promises of large language models for protein design and modeling. Valentini et al. (2023). Frontiers in Bioinformatics. doi:10.3389/fbinf.2023.1304099

  Abstract

  The recent breakthroughs of Large Language Models (LLMs) in the context of natural language processing have opened the way to significant advances in protein research. Indeed, the relationships between human natural language and the "language of proteins" invite the application and adaptation of LLMs to protein modelling and design. Considering the impressive results of GPT-4 and other recently developed LLMs in processing, generating and translating human languages, we anticipate analogous results with the language of proteins. Indeed, protein language models have been already trained to accurately predict protein properties, generate novel functionally characterized proteins, achieving state-of-the-art results. In this paper we discuss the promises and the open challenges raised by this novel and exciting research area, and we propose our perspective on how LLMs will affect protein modeling and design.

• Ofer et al. (2021) — Explains the analogy between natural language processing and protein sequences, from word embeddings to transformers. Link only because of the ND licence. (review; CC BY-NC-ND 4.0 (link only))

  The language of proteins: NLP, machine learning & protein sequences. Ofer et al. (2021). Computational and Structural Biotechnology Journal. doi:10.1016/j.csbj.2021.03.022

  Abstract

  Natural language processing (NLP) is a field of computer science concerned with automated text and language analysis. In recent years, following a series of breakthroughs in deep and machine learning, NLP methods have shown overwhelming progress. Here, we review the success, promise and pitfalls of applying NLP algorithms to the study of proteins. Proteins, which can be represented as strings of amino-acid letters, are a natural fit to many NLP methods. We explore the conceptual similarities and differences between proteins and language, and review a range of protein-related tasks amenable to machine learning. We present methods for encoding the information of proteins as text and analyzing it with NLP methods, reviewing classic concepts such as bag-of-words, k-mers/n-grams and text search, as well as modern techniques such as word embedding, contextualized embedding, deep learning and neural language models. In particular, we focus on recent innovations such as masked language modeling, self-supervised learning and attention-based models. Finally, we discuss trends and challenges in the intersection of NLP and protein research.

D.1.9 Benchmarking search methods

• Weber et al. (2019) — General guidelines for designing fair benchmarks: choosing methods and reference datasets, defining the ground truth and choosing evaluation metrics. These apply directly to the benchmarking of homology search methods. (review; CC BY 4.0)

  Essential guidelines for computational method benchmarking. Weber et al. (2019). Genome Biology. doi:10.1186/s13059-019-1738-8

  Abstract

  In computational biology and other sciences, researchers are frequently faced with a choice between several computational methods for performing data analyses. Benchmarking studies aim to rigorously compare the performance of different methods using well-characterized benchmark datasets, to determine the strengths of each method or to provide recommendations regarding suitable choices of methods for an analysis. However, benchmarking studies must be carefully designed and implemented to provide accurate, unbiased, and informative results. Here, we summarize key practical guidelines and recommendations for performing high-quality benchmarking analyses, based on our experiences in computational biology.

D.1.10 Whole-genome alignment and conservation

• Armstrong et al. (2019) — Explains pairwise and multiple whole-genome alignment, the Cactus approach, and how alignments are used to annotate genes and conserved elements across species. (review; licence not stated)

  Whole-Genome Alignment and Comparative Annotation. Armstrong et al. (2019). Annual Review of Animal Biosciences. doi:10.1146/annurev-animal-020518-115005

  Abstract

  Rapidly improving sequencing technology coupled with computational developments in sequence assembly are making reference-quality genome assembly economical. Hundreds of vertebrate genome assemblies are now publicly available, and projects are being proposed to sequence thousands of additional species in the next few years. Such dense sampling of the tree of life should give an unprecedented new understanding of evolution and allow a detailed determination of the events that led to the wealth of biodiversity around us. To gain this knowledge, these new genomes must be compared through genome alignment (at the sequence level) and comparative annotation (at the gene level). However, different alignment and annotation methods have different characteristics; before starting a comparative genomics analysis, it is important to understand the nature of, and biases and limitations inherent in, the chosen methods. This review is intended to act as a technical but high-level overview of the field that should provide this understanding. We briefly survey the state of the genome alignment and comparative annotation fields and potential future directions for these fields in a new, large-scale era of comparative genomics.

D.2 Databases

Further explanations for Chapter 2.

D.2.1 How biological databases work

• Oza et al. (2023) — Practical advice on finding, judging, downloading and citing data from public databases, including metadata, versions and access restrictions. A short, broad introduction to the issues discussed in this section. (tutorial; CC BY 4.0)

  Ten simple rules for using public biological data for your research. Oza et al. (2023). PLOS Computational Biology. doi:10.1371/journal.pcbi.1010749

  Abstract

  With an increasing amount of biological data available publicly, there is a need for a guide on how to successfully download and use this data. The 10 simple rules for using public biological data are: (1) use public data purposefully in your research; (2) evaluate data for your use case; (3) check data reuse requirements and embargoes; (4) be aware of ethics for data reuse; (5) plan for data storage and compute requirements; (6) know what you are downloading; (7) download programmatically and verify integrity; (8) properly cite data; (9) make reprocessed data and models Findable, Accessible, Interoperable, and Reusable (FAIR) and share; and (10) make pipelines and code FAIR and share. These rules are intended as a guide for researchers wanting to make use of available data and to increase data reuse and reproducibility.

• Danielewski et al. (2025) — Reviews why and how biological sequence databases were established, the move to open access, and current policy questions such as privacy and the international debate on digital sequence information. Useful background for the subsection on licences and access. (review; CC BY 4.0)

  History of Biological Databases, Their Importance, and Existence in Modern Scientific and Policy Context. Danielewski et al. (2025). Genes. doi:10.3390/genes16010100

  Abstract

  With the development of genome sequencing technologies, the amount of data produced has greatly increased in the last two decades. The abundance of digital sequence information (DSI) has provided research opportunities, improved our understanding of the genome, and led to the discovery of new solutions in industry and medicine. It has also posed certain challenges, i.e., how to store and handle such amounts of data. This, coupled with the need for convenience, international cooperation, and the possibility of independent validation, has led to the establishment of numerous databases. Spearheaded with the idea that data obtained with public funds should be available to the public, open access has become the predominant mode of accession. However, the increasing popularity of commercial genetic tests brings back the topic of data misuse, and patient's privacy. At the previous United Nations Biodiversity Conference (COP15, 2022), an issue of the least-developed countries exploiting their natural resources while providing DSI and the most-developed countries benefitting from this was raised. It has been proposed that financial renumeration for the data could help protect biodiversity. With the goal of introducing the topic to those interested in utilizing biological databases, in this publication, we present the history behind the biological databases, their necessity in today's scientific world, and the issues that concern them and their content, while providing scientific and policy context in relation to United Nations Biodiversity Conference (COP16, 21.10-1.11.24).

• Rosonovski et al. (2023) — Step-by-step protocols for searching the literature, full texts and preprints in Europe PMC, including programmatic access. Complements the subsection on PubMed, PMC and preprints. (tutorial; CC BY 4.0)

  Searching and Evaluating Publications and Preprints Using Europe PMC. Rosonovski et al. (2023). Current Protocols. doi:10.1002/cpz1.694

  Abstract

  In the field of life sciences there is a growing need for literature analysis tools that help scientists tackle information overload. Europe PubMed Central (Europe PMC), a partner of PubMed Central (PMC; National Library of Medicine, 2022), is an open access database of over 41 million life science publications and preprints, enriched with supporting data, reviews, protocols, and other relevant resources. Europe PMC is a trusted repository of choice for many life science funders (Europe PMC, 2022a), offering a suite of innovative search tools that allow users to search and evaluate the literature, including finding highly cited articles, preprints with community peer reviews, or papers referencing a proteomics dataset in the figure legend. In addition, Europe PMC utilizes text-mining to help researchers identify key terms and find data and evidence in the literature. First-time users often do not utilize the wealth of tools Europe PMC offers and can feel overwhelmed about how to perform the most effective search. This protocol, describing how to search and evaluate publications and preprints using Europe PMC, demonstrates how to carry out more efficient and effective literature searches using the tools provided by Europe PMC. This includes discovering the latest findings on a research topic, following research from a specific author, journal, or preprint server, exploring literature on a new method, expanding your reading list with relevant articles, as well as accessing and evaluating publications and preprints of interest. © 2023 EMBL-EBI. Current Protocols published by Wiley Periodicals LLC. Basic Protocol 1: Finding articles and preprints on a topic of interest Basic Protocol 2: Accessing an article Basic Protocol 3: Browsing the article Basic Protocol 4: Evaluating the article Basic Protocol 5: Refining search results Basic Protocol 6: Finding research by author Basic Protocol 7: Finding a specific article Basic Protocol 8: Finding information about a methodology Basic Protocol 9: Finding evidence of biological interactions, relations, and modifications Basic Protocol 10: Finding data behind a publication Basic Protocol 11: Expanding a reading list and building a bibliography Basic Protocol 12: Staying on top of the current literature.

D.2.2 Sequence databases

• Lussi et al. (2023) — Protocols written by the UniProt team on navigating and searching the UniProt datasets (UniProtKB, UniRef, UniParc and Proteomes) and retrieving data. A practical complement to the description of UniProt in this section. (tutorial; CC BY 4.0)

  Searching and Navigating UniProt Databases. Lussi et al. (2023). Current Protocols. doi:10.1002/cpz1.700

  Abstract

  The Universal Protein Resource (UniProt) is a comprehensive resource for protein sequence and annotation data. The UniProt website receives about 800,000 unique visitors per month and is the primary means to access UniProt. It provides 10 searchable datasets and four main tools. The key UniProt datasets are the UniProt Knowledgebase (UniProtKB), the UniProt Reference Clusters (UniRef), the UniProt Archive (UniParc), and protein sets for completely sequenced genomes (Proteomes). Other supporting datasets include information about proteins that is present in UniProtKB protein entries, such as literature citations, taxonomy, and subcellular locations, among others. This article focuses on how to use UniProt datasets. The first basic protocol describes navigation and searching mechanisms for the UniProt datasets, and two additional protocols build on the first protocol to describe advanced search and query building. © 2023 The Authors. Current Protocols published by Wiley Periodicals LLC. Basic Protocol 1: Searching UniProt datasets Basic Protocol 2: Advanced search and query building Basis Protocol 3: Adding parameters using advanced search.

• Vuong et al. (2022) — Guidance on working with environmental metagenomic data, including sample metadata, reference databases and data deposition. Relevant to the subsection on metagenomic sequences. (tutorial; CC BY 4.0)

  Ten simple rules for investigating (meta)genomic data from environmental ecosystems. Vuong et al. (2022). PLOS Computational Biology. doi:10.1371/journal.pcbi.1010675

  Abstract

  Metagenomics is the sequencing and study of DNA directly from environmental samples and is an approach used in microbial ecology to explore the diversity and functions of microbial communities [1].The study of the metagenome presents an avenue to investigate environmental microorganisms in-situ through culture-independent techniques, which facilitates the discovery and analysis of the metabolic potential of hitherto uncultured taxa [2].Expanding our reach into the once obscure facets of microbial life has yielded further insights into the ecological activity of microorganisms, uncovering a trove of novel metabolic products and pathways with a wide range of biotechnical applications in many industries [3,4].Data obtained from metagenomes have also been pivotal in directing strategies for the cultivation of novel microorganisms [5].The benefit of metagenomics is that it forms a wide-reaching preliminary approach that can provide contextual information to direct further downstream research for microorganisms in the wet lab.Microbial ecology has developed to a stage where "big data" is often the norm.Through advances in sequencing and computational technologies, we have experienced an unprecedented level of progress in the field of microbial ecology, improving our understanding of the many global ecosystems [6].Metagenomics draws many parallels when compared to data science-environmental sequence data presents a large volume of information that must be sifted through and sorted to make sense of underlying ecological and taxonomic patterns [7].The amount of information that needs to be processed, however, can be daunting for those starting out and often still pose challenges even for those familiar to the field.The inspiration behind this writing is from a recommendation to a PLOS Ten Simple Rules article on command-line bioinformatics [8].The article provided a lot of information on tools and concepts used by bioinformaticians and it would have been a valuable source of guidance for many.This paper styled in the typical format of a Ten Simple Rules article, presents a guide on the basics of metagenomic-based research aimed at environmental settings.We will explore the concepts, tools, and resources utilized in metagenomics and how these can potentially help the progress of a researcher beginning their studies (Fig 1).The 10 simple rules format is a casual one and people are often busy, so at the end of each rule is a TL; DR (too long; didn't read) that summarizes the point we are trying to make if you want to skim through (although we do hope you'll stick around and read the paper in depth!).

D.2.3 Genome databases and browsers

• Spudich and Fernández-Suárez (2010) — A worked tour of the Ensembl browser through four examples covering gene pages, comparative genomics, disease-linked variation and functional genomics. The interface has changed since 2010, but the concepts it explains still apply. (tutorial; CC BY (Europe PMC))

  Touring Ensembl: A practical guide to genome browsing. Spudich & Fernández-Suárez (2010). BMC Genomics. doi:10.1186/1471-2164-11-295

  Abstract

  The number of databases in molecular biological fields has rapidly increased to provide a large-scale resource. Though valuable information is available, data can be difficult to access, compare and integrate due to different formats and presentations of web interfaces. This paper offers a practical guide to the integration of gene, comparative genomic, and functional genomics data using the Ensembl website at http://www.ensembl.org.The Ensembl genome browser and underlying databases focus on chordate organisms. More species such as plants and microorganisms can be investigated using our sister browser at http://www.ensemblgenomes.org.In this study, four examples are used that sample many pages and features of the Ensembl browser. We focus on comparative studies across over 50 mostly chordate organisms, variations linked to disease, functional genomics, and access of external information housed in databases outside the Ensembl project. Researchers will learn how to go beyond simply exporting one gene sequence, and explore how a genome browser can integrate data from various sources and databases to build a full and comprehensive biological picture.

• Oliver et al. (2016) — Explains what model organism databases do, why expert curation of the literature matters, and the problem of funding them. Background for the subsection on organism-specific databases. (perspective; CC BY (Europe PMC))

  Model organism databases: essential resources that need the support of both funders and users. Oliver et al. (2016). BMC Biology. doi:10.1186/s12915-016-0276-z

  Abstract

  Modern biomedical research depends critically on access to databases that house and disseminate genetic, genomic, molecular, and cell biological knowledge. Even as the explosion of available genome sequences and associated genome-scale data continues apace, the sustainability of professionally maintained biological databases is under threat due to policy changes by major funding agencies. Here, we focus on model organism databases to demonstrate the myriad ways in which biological databases not only act as repositories but actively facilitate advances in research. We present data that show that reducing financial support to model organism databases could prove to be not just scientifically, but also economically, unsound.

• Dominguez Del Angel et al. (2018) — An overview of how a genome is assembled and annotated before it appears in genome databases. Helps to understand where gene models and assembly versions come from. (tutorial; CC BY 4.0)

  Ten steps to get started in Genome Assembly and Annotation. Dominguez Del Angel et al. (2018). F1000Research. doi:10.12688/f1000research.13598.1

  Abstract

  As a part of the ELIXIR-EXCELERATE efforts in capacity building, we present here 10 steps to facilitate researchers getting started in genome assembly and genome annotation. The guidelines given are broadly applicable, intended to be stable over time, and cover all aspects from start to finish of a general assembly and annotation project. Intrinsic properties of genomes are discussed, as is the importance of using high quality DNA. Different sequencing technologies and generally applicable workflows for genome assembly are also detailed. We cover structural and functional annotation and encourage readers to also annotate transposable elements, something that is often omitted from annotation workflows. The importance of data management is stressed, and we give advice on where to submit data and how to make your results Findable, Accessible, Interoperable, and Reusable (FAIR).

D.2.4 Structure databases

• David et al. (2022) — Explains how to use and interpret predicted structures from the AlphaFold Database, including the pLDDT and PAE confidence measures. Useful for the subsection on model databases. (review; CC BY (Europe PMC))

  The AlphaFold Database of Protein Structures: A Biologist’s Guide. David et al. (2022). Journal of Molecular Biology. doi:10.1016/j.jmb.2021.167336

  Abstract

  AlphaFold, the deep learning algorithm developed by DeepMind, recently released the three-dimensional models of the whole human proteome to the scientific community. Here we discuss the advantages, limitations and the still unsolved challenges of the AlphaFold models from the perspective of a biologist, who may not be an expert in structural biology.

• Burley et al. (2024) — Describes the PDB archive, its deposition and validation, and the RCSB tools for searching and visualising both experimental structures and computed models. (review; CC BY 4.0)

  RCSB Protein Data Bank: supporting research and education worldwide through explorations of experimentally determined and computationally predicted atomic level 3D biostructures. Burley et al. (2024). IUCrJ. doi:10.1107/s2052252524002604

  Abstract

  The Protein Data Bank (PDB) was established as the first open-access digital data resource in biology and medicine in 1971 with seven X-ray crystal structures of proteins. Today, the PDB houses >210 000 experimentally determined, atomic level, 3D structures of proteins and nucleic acids as well as their complexes with one another and small molecules (e.g. approved drugs, enzyme cofactors). These data provide insights into fundamental biology, biomedicine, bioenergy and biotechnology. They proved particularly important for understanding the SARS-CoV-2 global pandemic. The US-funded Research Collaboratory for Structural Bioinformatics Protein Data Bank (RCSB PDB) and other members of the Worldwide Protein Data Bank (wwPDB) partnership jointly manage the PDB archive and support >60 000 `data depositors' (structural biologists) around the world. wwPDB ensures the quality and integrity of the data in the ever-expanding PDB archive and supports global open access without limitations on data usage. The RCSB PDB research-focused web portal at https://www.rcsb.org/ (RCSB.org) supports millions of users worldwide, representing a broad range of expertise and interests. In addition to retrieving 3D structure data, PDB `data consumers' access comparative data and external annotations, such as information about disease-causing point mutations and genetic variations. RCSB.org also provides access to >1 000 000 computed structure models (CSMs) generated using artificial intelligence/machine-learning methods. To avoid doubt, the provenance and reliability of experimentally determined PDB structures and CSMs are identified. Related training materials are available to support users in their RCSB.org explorations.

• Varadi et al. (2022) — Describes the European PDB node and its knowledge base, which links structures to sequence, function and ligand annotation. Shows how structure data are integrated with other databases. (review; CC BY 4.0)

  PDBe and PDBe-KB: Providing high-quality, up-to-date and integrated resources of macromolecular structures to support basic and applied research and education. Varadi et al. (2022). Protein Science. doi:10.1002/pro.4439

  Abstract

  The archiving and dissemination of protein and nucleic acid structures as well as their structural, functional and biophysical annotations is an essential task that enables the broader scientific community to conduct impactful research in multiple fields of the life sciences. The Protein Data Bank in Europe (PDBe; pdbe.org) team develops and maintains several databases and web services to address this fundamental need. From data archiving as a member of the Worldwide PDB consortium (wwPDB; wwpdb.org), to the PDBe Knowledge Base (PDBe-KB; pdbekb.org), we provide data, data-access mechanisms, and visualizations that facilitate basic and applied research and education across the life sciences. Here, we provide an overview of the structural data and annotations that we integrate and make freely available. We describe the web services and data visualization tools we offer, and provide information on how to effectively use or even further develop them. Finally, we discuss the direction of our data services, and how we aim to tackle new challenges that arise from the recent, unprecedented advances in the field of structure determination and protein structure modeling.

D.2.5 Function and pathways

• Gaudet et al. (2016) — A short introduction to the structure of GO, its three aspects, relations between terms and how annotations with evidence codes are made. A good companion to the Gene Ontology subsection. (book chapter; CC BY 4.0)

  Primer on the Gene Ontology. Gaudet et al. (2016). The Gene Ontology Handbook. doi:10.1007/978-1-4939-3743-1_3

  Abstract

  The Gene Ontology (GO) project is the largest resource for cataloguing gene function. The combination of solid conceptual underpinnings and a practical set of features have made the GO a widely adopted resource in the research community and an essential resource for data analysis. In this chapter, we provide a concise primer for all users of the GO. We briefly introduce the structure of the ontology and explain how to interpret annotations associated with the GO.

• Rothfels et al. (2023) — Protocols from the Reactome team on browsing pathways, interpreting the pathway diagrams and running pathway analysis on a gene list. (tutorial; CC BY 4.0)

  Using the Reactome Database. Rothfels et al. (2023). Current Protocols. doi:10.1002/cpz1.722

  Abstract

  Pathway databases provide descriptions of the roles of proteins, nucleic acids, lipids, carbohydrates, and other molecular entities within their biological cellular contexts. Pathway-centric views of these roles may allow for the discovery of unexpected functional relationships in data such as gene expression profiles and somatic mutation catalogues from tumor cells. For this reason, there is a high demand for high-quality pathway databases and their associated tools. The Reactome project (a collaboration between the Ontario Institute for Cancer Research, New York University Langone Health, the European Bioinformatics Institute, and Oregon Health & Science University) is one such pathway database. Reactome collects detailed information on biological pathways and processes in humans from the primary literature. Reactome content is manually curated, expert-authored, and peer-reviewed and spans the gamut from simple intermediate metabolism to signaling pathways and complex cellular events. This information is supplemented with likely orthologous molecular reactions in mouse, rat, zebrafish, worm, and other model organisms. © 2023 The Authors. Current Protocols published by Wiley Periodicals LLC. Basic Protocol 1: Browsing a Reactome pathway Basic Protocol 2: Exploring Reactome annotations of disease and drugs Basic Protocol 3: Finding the pathways involving a gene or protein Alternate Protocol 1: Finding the pathways involving a gene or protein using UniProtKB (SwissProt), Ensembl, or Entrez gene identifier Alternate Protocol 2: Using advanced search Basic Protocol 4: Using the Reactome pathway analysis tool to identify statistically overrepresented pathways Basic Protocol 5: Using the Reactome pathway analysis tool to overlay expression data onto Reactome pathway diagrams Basic Protocol 6: Comparing inferred model organism and human pathways using the Species Comparison tool Basic Protocol 7: Comparing tissue-specific expression using the Tissue Distribution tool.

• Chicco and Agapito (2022) — Practical advice on using GO and pathway databases for enrichment analysis, including choice of background and database version. Shows common pitfalls when function databases are used for interpretation. (tutorial; CC BY 4.0)

  Nine quick tips for pathway enrichment analysis. Chicco & Agapito (2022). PLOS Computational Biology. doi:10.1371/journal.pcbi.1010348

  Abstract

  Pathway enrichment analysis (PEA) is a computational biology method that identifies biological functions that are overrepresented in a group of genes more than would be expected by chance and ranks these functions by relevance. The relative abundance of genes pertinent to specific pathways is measured through statistical methods, and associated functional pathways are retrieved from online bioinformatics databases. In the last decade, along with the spread of the internet, higher availability of computational resources made PEA software tools easy to access and to use for bioinformatics practitioners worldwide. Although it became easier to use these tools, it also became easier to make mistakes that could generate inflated or misleading results, especially for beginners and inexperienced computational biologists. With this article, we propose nine quick tips to avoid common mistakes and to out a complete, sound, thorough PEA, which can produce relevant and robust results. We describe our nine guidelines in a simple way, so that they can be understood and used by anyone, including students and beginners. Some tips explain what to do before starting a PEA, others are suggestions of how to correctly generate meaningful results, and some final guidelines indicate some useful steps to properly interpret PEA results. Our nine tips can help users perform better pathway enrichment analyses and eventually contribute to a better understanding of current biology.

D.2.6 Interaction databases

• De Las Rivas and Fontanillo (2010) — Explains the types of protein interaction data, the experimental methods behind them and the primary and integrated interaction databases. Older, but clear on the distinction between physical and functional interactions. (tutorial; CC BY 4.0)

  Protein–Protein Interactions Essentials: Key Concepts to Building and Analyzing Interactome Networks. De Las Rivas & Fontanillo (2010). PLoS Computational Biology. doi:10.1371/journal.pcbi.1000807

• Panneerselvam et al. (2024) — Protocols for searching IntAct and using the detailed curation of experimental context that IMEx databases record. Illustrates what a curated physical interaction record contains. (tutorial; CC BY 4.0)

  IntAct Database for Accessing IMEx’s Contextual Metadata of Molecular Interactions. Panneerselvam et al. (2024). Current Protocols. doi:10.1002/cpz1.70018

  Abstract

  The International Molecular Exchange Consortium (IMEx) has evolved into a vital partnership of open resources dedicated to curating molecular interaction data from the scientific literature. This consortium, which includes IntAct, MINT, MatrixDB, and DIP, is a collaborative effort with a central mission of aggregating detailed molecular interaction experimental evidence in a machine-readable format, supported by controlled vocabularies and standard ontologies. The IntAct molecular interaction database (www.ebi.ac.uk/intact), as an IMEx partner, serves as a valuable portal for accessing IMEx data through user-friendly search options and an array of interactive filters. The resource currently hosts an extensive repository of 1,293,508 binary interactions meticulously captured from 75,098 experiments documented in 23,366 publications (as of the February 2024 release), with this corpora being added to by regular data releases. IMEx curation policy has consistently prioritized a fine-grained data and curation model, with a focus on capturing the relevant experimental details essential for interpreting molecular interaction data effectively. Our curation process is designed to support the generation of interactomes tailored to contexts such as disease-specific or tissue-/cell-type-specific interactomes. These interactions are ranked according to a scoring system based on the Proteomics Standard Initiative Molecular Interaction (PSI MI) standards. This scoring system allows users to assess the degree of confidence in binary interactions, enhancing the value of the data. The resource provides insights into the nature of relationships among interacting partners as defined by the experimental setup and the associated biological context. Interactive filters enable users to navigate these rich, multilayered data, promoting a deeper understanding of biological complexity. Additionally, the IntAct website fosters the creation of networks for collaborative analyses by the scientific community. The recent transformation of the IntAct website, supported by a graph-type database, empowers users to execute custom queries tailored to their specific research interests. This article illustrates the diverse levels of annotations available for interactions and the multiple search options at users' disposal to access data of interest. © 2024 European Molecular Biology Laboratory, European Bioinformatics Institute. Current Protocols published by Wiley Periodicals LLC. Basic Protocol 1: Using Quick Search, network visualization, and filters Support Protocol: Accessing fine annotations from intact: Unlocking the molecular details Alternate Protocol: Using batch search: Querying multiple interactors Basic Protocol 2: Using advanced search: Precision and customization.

D.2.7 Expression and proteomics repositories

• Wang et al. (2018) — Reviews methods for curating and reprocessing GEO datasets at scale, and explains how poorly standardised metadata limits reuse. Useful for understanding both the value and the limitations of expression repositories. (review; CC BY 4.0)

  Mining data and metadata from the gene expression omnibus. Wang et al. (2018). Biophysical Reviews. doi:10.1007/s12551-018-0490-8

  Abstract

  Publicly available gene expression datasets deposited in the Gene Expression Omnibus (GEO) are growing at an accelerating rate. Such datasets hold great value for knowledge discovery, particularly when integrated. Although numerous software platforms and tools have been developed to enable reanalysis and integration of individual, or groups, of GEO datasets, large-scale reuse of those datasets is impeded by minimal requirements for standardized metadata both at the study and sample levels as well as uniform processing of the data across studies. Here, we review methodologies developed to facilitate the systematic curation and processing of publicly available gene expression datasets from GEO. We identify trends for advanced metadata curation and summarize approaches for reprocessing the data within the entire GEO repository.

• Perez-Riverol et al. (2015) — An overview of proteomics repositories, ProteomeXchange and data standards. Some details are dated, but it explains the landscape that PRIDE and related resources belong to. (review; CC BY 4.0)

  Making proteomics data accessible and reusable: Current state of proteomics databases and repositories. Perez-Riverol et al. (2015). PROTEOMICS. doi:10.1002/pmic.201400302

  Abstract

  Compared to other data-intensive disciplines such as genomics, public deposition and storage of MS-based proteomics, data are still less developed due to, among other reasons, the inherent complexity of the data and the variety of data types and experimental workflows. In order to address this need, several public repositories for MS proteomics experiments have been developed, each with different purposes in mind. The most established resources are the Global Proteome Machine Database (GPMDB), PeptideAtlas, and the PRIDE database. Additionally, there are other useful (in many cases recently developed) resources such as ProteomicsDB, Mass Spectrometry Interactive Virtual Environment (MassIVE), Chorus, MaxQB, PeptideAtlas SRM Experiment Library (PASSEL), Model Organism Protein Expression Database (MOPED), and the Human Proteinpedia. In addition, the ProteomeXchange consortium has been recently developed to enable better integration of public repositories and the coordinated sharing of proteomics information, maximizing its benefit to the scientific community. Here, we will review each of the major proteomics resources independently and some tools that enable the integration, mining and reuse of the data. We will also discuss some of the major challenges and current pitfalls in the integration and sharing of the data.

D.2.8 Medical and variant data

• Riccio et al. (2024) — A systematic review of 118 databases and software tools for annotating the functional impact of genetic variants, with practical advice on which to combine. Relevant to the subsection on variants and their interpretation. (review; CC BY 4.0)

  Variant effect predictors: a systematic review and practical guide. Riccio et al. (2024). Human Genetics. doi:10.1007/s00439-024-02670-5

  Abstract

  Large-scale association analyses using whole-genome sequence data have become feasible, but understanding the functional impacts of these associations remains challenging. Although many tools are available to predict the functional impacts of genetic variants, it is unclear which tool should be used in practice. This work provides a practical guide to assist in selecting appropriate tools for variant annotation. We conducted a MEDLINE search up to November 10, 2023, and included tools that are applicable to a broad range of phenotypes, can be used locally, and have been recently updated. Tools were categorized based on the types of variants they accept and the functional impacts they predict. Sequence Ontology terms were used for standardization. We identified 118 databases and software packages, encompassing 36 variant types and 161 functional impacts. Combining only three tools, namely SnpEff, FAVOR, and SparkINFERNO, allows predicting 99 (61%) distinct functional impacts. Thirty-seven tools predict 89 functional impacts that are not supported by any other tool, while 75 tools predict pathogenicity and can be used within the ACMG/AMP guidelines in a clinical context. We launched a website allowing researchers to select tools based on desired variants and impacts. In summary, more than 100 tools are already available to predict approximately 160 functional impacts. About 60% of the functional impacts can be predicted by the combination of three tools. Unexpectedly, recent tools do not predict more impacts than older ones. Future research should allow predicting the functionality of so far unsupported variant types, such as gene fusions.URL: https://cardio-care.shinyapps.io/VEP_Finder/ .Registration: OSF Registries on November 10, 2023, https://osf.io/s2gct .

• Corpas et al. (2018) — Guidelines for providers of human genomic data on making shared datasets findable, accessible, interoperable and reusable. Background for the subsection on controlled access to personal data. (tutorial; CC BY 4.0)

  A FAIR guide for data providers to maximise sharing of human genomic data. Corpas et al. (2018). PLOS Computational Biology. doi:10.1371/journal.pcbi.1005873

  Abstract

  It is generally acknowledged that, for reproducibility and progress of human genomic research, data sharing is critical. For every sharing transaction, a successful data exchange is produced between a data consumer and a data provider. Providers of human genomic data (e.g., publicly or privately funded repositories and data archives) fulfil their social contract with data donors when their shareable data conforms to FAIR (findable, accessible, interoperable, reusable) principles. Based on our experiences via Repositive (https://repositive.io), a leading discovery platform cataloguing all shared human genomic datasets, we propose guidelines for data providers wishing to maximise their shared data's FAIRness.

• Van Buren et al. (2025) — Compares human short-variant databases with those built for animal species, covering methods, data integration and applications. Gives a wider view of how variant repositories are organised. (review; CC BY-NC 4.0 (link only))

  A comparative review of short genetic variant databases across humans and animal species. Van Buren et al. (2025). Briefings in Bioinformatics. doi:10.1093/bib/bbaf356

  Abstract

  Understanding genetic variation is necessary to unravel the complexities of evolution and diverse traits across species. Short genetic variants (<50 bp in length) represent key genomic variations that play a crucial role in shaping the genetic landscape of populations. Human short genetic variant databases are often considered the gold standard for variant repositories, and this review compares them with variant databases created for various animal species. This review examines the methodologies, data integration, and various applications that differ or align between human and animal databases. The goal is to identify challenges in leveraging genetic variation across species, recommend strategies for overcoming these challenges, and suggest future directions for research and database development.

D.3 Evolution

Further explanations for Chapter 3.

D.3.1 The origin of life

• Preiner et al. (2020) — Surveys the main competing scenarios (RNA world, metabolism first, compartments, vent and surface settings) and argues for combining them. A useful overview of the field as it stands after the classic debates. (review; CC BY 4.0)

  The Future of Origin of Life Research: Bridging Decades-Old Divisions. Preiner et al. (2020). Life. doi:10.3390/life10030020

  Abstract

  Research on the origin of life is highly heterogeneous. After a peculiar historical development, it still includes strongly opposed views which potentially hinder progress. In the 1st Interdisciplinary Origin of Life Meeting, early-career researchers gathered to explore the commonalities between theories and approaches, critical divergence points, and expectations for the future. We find that even though classical approaches and theories-e.g. bottom-up and top-down, RNA world vs. metabolism-first-have been prevalent in origin of life research, they are ceasing to be mutually exclusive and they can and should feed integrating approaches. Here we focus on pressing questions and recent developments that bridge the classical disciplines and approaches, and highlight expectations for future endeavours in origin of life research.

• Pross and Pascal (2013) — A short, readable essay on which questions about the origin of life are answerable in principle and which are probably not, including the historical uncertainty of any scenario. (perspective; CC BY (Europe PMC))

  The origin of life: what we know, what we can know and what we will never know. Pross & Pascal (2013). Open Biology. doi:10.1098/rsob.120190

  Abstract

  The origin of life (OOL) problem remains one of the more challenging scientific questions of all time. In this essay, we propose that following recent experimental and theoretical advances in systems chemistry, the underlying principle governing the emergence of life on the Earth can in its broadest sense be specified, and may be stated as follows: all stable (persistent) replicating systems will tend to evolve over time towards systems of greater stability. The stability kind referred to, however, is dynamic kinetic stability, and quite distinct from the traditional thermodynamic stability which conventionally dominates physical and chemical thinking. Significantly, that stability kind is generally found to be enhanced by increasing complexification, since added features in the replicating system that improve replication efficiency will be reproduced, thereby offering an explanation for the emergence of life's extraordinary complexity. On the basis of that simple principle, a fundamental reassessment of the underlying chemistry-biology relationship is possible, one with broad ramifications. In the context of the OOL question, this novel perspective can assist in clarifying central ahistoric aspects of abiogenesis, as opposed to the many historic aspects that have probably been forever lost in the mists of time.

D.3.2 Basic evolutionary concepts

• Aris-Brosou and Rodrigue (2019) — An introduction to the population-genetic and probabilistic foundations of molecular evolution, from mutation, selection and drift to substitution models. Bridges this section and the methods in the phylogeny chapter. (book chapter; CC BY 4.0)

  A Not-So-Long Introduction to Computational Molecular Evolution. Aris-Brosou & Rodrigue (2019). Evolutionary Genomics. doi:10.1007/978-1-4939-9074-0_3

  Abstract

  In this chapter, we give a not-so-long and self-contained introduction to computational molecular evolution. In particular, we present the emergence of the use of likelihood-based methods, review the standard DNA substitution models, and introduce how model choice operates. We also present recent developments in inferring absolute divergence times and rates on a phylogeny, before showing how state-of-the-art models take inspiration from diffusion theory to link population genetics, which traditionally focuses at a taxonomic level below that of the species, and molecular evolution. Although this is not a cookbook chapter, we try and point to popular programs and implementations along the way.

• Casillas and Barbadilla (2017) — A historical and conceptual review of molecular population genetics, covering the neutral and nearly neutral theories, linked selection and the interplay of drift and selection in genomic data. (review; CC BY (Europe PMC))

  Molecular Population Genetics. Casillas & Barbadilla (2017). Genetics. doi:10.1534/genetics.116.196493

  Abstract

  Molecular population genetics aims to explain genetic variation and molecular evolution from population genetics principles. The field was born 50 years ago with the first measures of genetic variation in allozyme loci, continued with the nucleotide sequencing era, and is currently in the era of population genomics. During this period, molecular population genetics has been revolutionized by progress in data acquisition and theoretical developments. The conceptual elegance of the neutral theory of molecular evolution or the footprint carved by natural selection on the patterns of genetic variation are two examples of the vast number of inspiring findings of population genetics research. Since the inception of the field, Drosophila has been the prominent model species: molecular variation in populations was first described in Drosophila and most of the population genetics hypotheses were tested in Drosophila species. In this review, we describe the main concepts, methods, and landmarks of molecular population genetics, using the Drosophila model as a reference. We describe the different genetic data sets made available by advances in molecular technologies, and the theoretical developments fostered by these data. Finally, we review the results and new insights provided by the population genomics approach, and conclude by enumerating challenges and new lines of inquiry posed by increasingly large population scale sequence data.

D.3.3 Homologues, orthologues and gene duplication

• Birchler and Yang (2022) — Reviews what happens to genes after small-scale and whole-genome duplication, including the dosage-balance hypothesis. Examples are mostly from plants, but the concepts are general. (review; CC BY 4.0)

  The multiple fates of gene duplications: Deletion, hypofunctionalization, subfunctionalization, neofunctionalization, dosage balance constraints, and neutral variation. Birchler & Yang (2022). The Plant Cell. doi:10.1093/plcell/koac076

  Abstract

  Gene duplications have long been recognized as a contributor to the evolution of genes with new functions. Multiple copies of genes can result from tandem duplication, from transposition to new chromosomes, or from whole-genome duplication (polyploidy). The most common fate is that one member of the pair is deleted to return the gene to the singleton state. Other paths involve the reduced expression of both copies (hypofunctionalization) that are held in duplicate to maintain sufficient quantity of function. The two copies can split functions (subfunctionalization) or can diverge to generate a new function (neofunctionalization). Retention of duplicates resulting from doubling of the whole genome occurs for genes involved with multicomponent interactions such as transcription factors and signal transduction components. In contrast, these classes of genes are underrepresented in small segmental duplications. This complementary pattern suggests that the balance of interactors affects the fate of the duplicate pair. We discuss the different mechanisms that maintain duplicated genes, which may change over time and intersect.

• Glover et al. (2019) — A community review from the Quest for Orthologs consortium on orthology concepts, inference methods, benchmarking and applications; complements the Altenhoff et al. chapter already cited. (review; CC BY 4.0)

  Advances and Applications in the Quest for Orthologs. Glover et al. (2019). Molecular Biology and Evolution. doi:10.1093/molbev/msz150

  Abstract

  Gene families evolve by the processes of speciation (creating orthologs), gene duplication (paralogs), and horizontal gene transfer (xenologs), in addition to sequence divergence and gene loss. Orthologs in particular play an essential role in comparative genomics and phylogenomic analyses. With the continued sequencing of organisms across the tree of life, the data are available to reconstruct the unique evolutionary histories of tens of thousands of gene families. Accurate reconstruction of these histories, however, is a challenging computational problem, and the focus of the Quest for Orthologs Consortium. We review the recent advances and outstanding challenges in this field, as revealed at a symposium and meeting held at the University of Southern California in 2017. Key advances have been made both at the level of orthology algorithm development and with respect to coordination across the community of algorithm developers and orthology end-users. Applications spanned a broad range, including gene function prediction, phylostratigraphy, genome evolution, and phylogenomics. The meetings highlighted the increasing use of meta-analyses integrating results from multiple different algorithms, and discussed ongoing challenges in orthology inference as well as the next steps toward improvement and integration of orthology resources.

• Kellis et al. (2016a) — Lecture-note style text on gene trees versus species trees and reconciliation to infer duplications, losses, orthologues and paralogues. (book chapter; CC BY-NC-SA (book level; page marked ‘not declared’) – link only)

  Inferring Orthologs/Paralogs, Gene Duplication and Loss. Kellis et al. (2016). Computational Biology: Genomes, Networks, Evolution (MIT 6.047, LibreTexts). https://bio.libretexts.org/Bookshelves/Computational_Biology/Book%3A_Computational_Biology_-_Genomes_Networks_and_Evolution_(Kellis_et_al.)/27%3A_Phylogenomics_II/27.05%3A_Inferring_Orthologs_Paralogs_Gene_Duplication_and_Loss

D.3.4 Protein evolution

• Forslund et al. (2019) — Describes how domain architectures arise by fusion, fission, duplication and loss of domains, and how they are analysed computationally; matches the domain-evolution subsection closely. (book chapter; CC BY 4.0)

  Evolution of Protein Domain Architectures. Forslund et al. (2019). Evolutionary Genomics. doi:10.1007/978-1-4939-9074-0_15

  Abstract

  This chapter reviews current research on how protein domain architectures evolve. We begin by summarizing work on the phylogenetic distribution of proteins, as this will directly impact which domain architectures can be formed in different species. Studies relating domain family size to occurrence have shown that they generally follow power law distributions, both within genomes and larger evolutionary groups. These findings were subsequently extended to multi-domain architectures. Genome evolution models that have been suggested to explain the shape of these distributions are reviewed, as well as evidence for selective pressure to expand certain domain families more than others. Each domain has an intrinsic combinatorial propensity, and the effects of this have been studied using measures of domain versatility or promiscuity. Next, we study the principles of protein domain architecture evolution and how these have been inferred from distributions of extant domain arrangements. Following this, we review inferences of ancestral domain architecture and the conclusions concerning domain architecture evolution mechanisms that can be drawn from these. Finally, we examine whether all known cases of a given domain architecture can be assumed to have a single common origin (monophyly) or have evolved convergently (polyphyly). We end by a discussion of some available tools for computational analysis or exploitation of protein domain architectures and their evolution.

• Joy et al. (2016) — An educational topic page on ancestral reconstruction with parsimony, maximum likelihood and Bayesian methods, including the resurrection of ancestral proteins. (tutorial; CC BY 4.0)

  Ancestral Reconstruction. Joy et al. (2016). PLOS Computational Biology. doi:10.1371/journal.pcbi.1004763

• Oliveira and Deane (2017) — A short review of how correlated mutations in multiple sequence alignments are used to predict residue contacts, protein structures and interactions. (review; CC BY 4.0)

  Co-evolution techniques are reshaping the way we do structural bioinformatics. de Oliveira & Deane (2017). F1000Research. doi:10.12688/f1000research.11543.1

  Abstract

  Co-evolution techniques were originally conceived to assist in protein structure prediction by inferring pairs of residues that share spatial proximity. However, the functional relationships that can be extrapolated from co-evolution have also proven to be useful in a wide array of structural bioinformatics applications. These techniques are a powerful way to extract structural and functional information in a sequence-rich world.

D.3.5 Detecting selection

• Kosiol and Anisimova (2019) — Overview of methods to detect selection from both divergence (codon models, dN/dS) and polymorphism data, with their assumptions. (book chapter; CC BY 4.0)

  Selection Acting on Genomes. Kosiol & Anisimova (2019). Evolutionary Genomics. doi:10.1007/978-1-4939-9074-0_12

  Abstract

  Populations evolve as mutations arise in individual organisms and, through hereditary transmission, may become "fixed" (shared by all individuals) in the population. Most mutations are lethal or have negative fitness consequences for the organism. Others have essentially no effect on organismal fitness and can become fixed through the neutral stochastic process known as random drift. However, mutations may also produce a selective advantage that boosts their chances of reaching fixation. Regions of genomes where new mutations are beneficial, rather than neutral or deleterious, tend to evolve more rapidly due to positive selection. Genes involved in immunity and defense are a well-known example; rapid evolution in these genes presumably occurs because new mutations help organisms to prevail in evolutionary "arms races" with pathogens. In recent years genome-wide scans for selection have enlarged our understanding of the genome evolution of various species. In this chapter, we will focus on methods to detect selection on the genome. In particular, we will discuss probabilistic models and how they have changed with the advent of new genome-wide data now available.

• Jones et al. (2019) — Discusses how model misspecification and data problems can produce false signals of positive selection in codon-model analyses; relevant to the pitfalls subsection. (book chapter; CC BY 4.0)

  Looking for Darwin in Genomic Sequences: Validity and Success Depends on the Relationship Between Model and Data. Jones et al. (2019). Evolutionary Genomics. doi:10.1007/978-1-4939-9074-0_13

  Abstract

  Codon substitution models (CSMs) are commonly used to infer the history of natural section for a set of protein-coding sequences, often with the explicit goal of detecting the signature of positive Darwinian selection. However, the validity and success of CSMs used in conjunction with the maximum likelihood (ML) framework is sometimes challenged with claims that the approach might too often support false conclusions. In this chapter, we use a case study approach to identify four legitimate statistical difficulties associated with inference of evolutionary events using CSMs. These include: (1) model misspecification, (2) low information content, (3) the confounding of processes, and (4) phenomenological load, or PL. While past criticisms of CSMs can be connected to these issues, the historical critiques were often misdirected, or overstated, because they failed to recognize that the success of any model-based approach depends on the relationship between model and data. Here, we explore this relationship and provide a candid assessment of the limitations of CSMs to extract historical information from extant sequences. To aid in this assessment, we provide a brief overview of: (1) a more realistic way of thinking about the process of codon evolution framed in terms of population genetic parameters, and (2) a novel presentation of the ML statistical framework. We then divide the development of CSMs into two broad phases of scientific activity and show that the latter phase is characterized by increases in model complexity that can sometimes negatively impact inference of evolutionary mechanisms. Such problems are not yet widely appreciated by the users of CSMs. These problems can be avoided by using a model that is appropriate for the data; but, understanding the relationship between the data and a fitted model is a difficult task. We argue that the only way to properly understand that relationship is to perform in silico experiments using a generating process that can mimic the data as closely as possible. The mutation-selection modeling framework (MutSel) is presented as the basis of such a generating process. We contend that if complex CSMs continue to be developed for testing explicit mechanistic hypotheses, then additional analyses such as those described in here (e.g., penalized LRTs and estimation of PL) will need to be applied alongside the more traditional inferential methods.

• Kellis et al. (2016b) — Lecture-note text on genomic signatures of selection: dN/dS, Tajima’s D, Fay and Wu’s H, FST and haplotype-based sweep tests, with human examples. (book chapter; CC BY-NC-SA (book level; page marked ‘not declared’) – link only)

  Natural Selection. Kellis et al. (2016). Computational Biology: Genomes, Networks, Evolution (MIT 6.047, LibreTexts). https://bio.libretexts.org/Bookshelves/Computational_Biology/Book%3A_Computational_Biology_-_Genomes_Networks_and_Evolution_(Kellis_et_al.)/29%3A_Population_Genetic_Variation/29.04%3A_Natural_Selection

D.3.6 Genome evolution

• Koonin (2016) — Reviews the extent and mechanisms of horizontal gene transfer in prokaryotes and argues that it is essential for their long-term survival and evolvability. (review; CC BY 4.0)

  Horizontal gene transfer: essentiality and evolvability in prokaryotes, and roles in evolutionary transitions. Koonin (2016). F1000Research. doi:10.12688/f1000research.8737.1

  Abstract

  The wide spread of gene exchange and loss in the prokaryotic world has prompted the concept of 'lateral genomics' to the point of an outright denial of the relevance of phylogenetic trees for evolution. However, the pronounced coherence congruence of the topologies of numerous gene trees, particularly those for (nearly) universal genes, translates into the notion of a statistical tree of life (STOL), which reflects a central trend of vertical evolution. The STOL can be employed as a framework for reconstruction of the evolutionary processes in the prokaryotic world. Quantitatively, however, horizontal gene transfer (HGT) dominates microbial evolution, with the rate of gene gain and loss being comparable to the rate of point mutations and much greater than the duplication rate. Theoretical models of evolution suggest that HGT is essential for the survival of microbial populations that otherwise deteriorate due to the Muller's ratchet effect. Apparently, at least some bacteria and archaea evolved dedicated vehicles for gene transfer that evolved from selfish elements such as plasmids and viruses. Recent phylogenomic analyses suggest that episodes of massive HGT were pivotal for the emergence of major groups of organisms such as multiple archaeal phyla as well as eukaryotes. Similar analyses appear to indicate that, in addition to donating hundreds of genes to the emerging eukaryotic lineage, mitochondrial endosymbiosis severely curtailed HGT. These results shed new light on the routes of evolutionary transitions, but caution is due given the inherent uncertainty of deep phylogenies.

• Puigbò et al. (2019) — Explains how comparing thousands of gene trees quantifies horizontal gene transfer and the remaining tree-like signal in prokaryotic evolution. (book chapter; CC BY 4.0)

  Genome-Wide Comparative Analysis of Phylogenetic Trees: The Prokaryotic Forest of Life. Puigbò et al. (2019). Evolutionary Genomics. doi:10.1007/978-1-4939-9074-0_8

  Abstract

  Genome-wide comparison of phylogenetic trees is becoming an increasingly common approach in evolutionary genomics, and a variety of approaches for such comparison have been developed. In this article we present several methods for comparative analysis of large numbers of phylogenetic trees. To compare phylogenetic trees taking into account the bootstrap support for each internal branch, the boot-split distance (BSD) method is introduced as an extension of the previously developed split distance (SD) method for tree comparison. The BSD method implements the straightforward idea that comparison of phylogenetic trees can be made more robust by treating tree splits differentially depending on the bootstrap support. Approaches are also introduced for detecting treelike and netlike evolutionary trends in the phylogenetic Forest of Life (FOL), i.e., the entirety of the phylogenetic trees for conserved genes of prokaryotes. The principal method employed for this purpose includes mapping quartets of species onto trees to calculate the support of each quartet topology and so to quantify the tree and net contributions to the distances between species. We describe the applications methods used to analyze the FOL and the results obtained with these methods. These results support the concept of the Tree of Life (TOL) as a central evolutionary trend in the FOL as opposed to the traditional view of the TOL as a "species tree."

D.4 Phylogeny

Further explanations for Chapter 4.

D.4.1 Basic definitions

• Jacques et al. (2023) — A step-by-step practical guide to gene and protein phylogenetics, from homologue collection and alignment to tree building and interpretation. Useful as a broad overview of the whole workflow behind the definitions in this section. (tutorial; CC BY 4.0)

  Roadmap to the study of gene and protein phylogeny and evolution—A practical guide. Jacques et al. (2023). PLOS ONE. doi:10.1371/journal.pone.0279597

  Abstract

  Developments in sequencing technologies and the sequencing of an ever-increasing number of genomes have revolutionised studies of biodiversity and organismal evolution. This accumulation of data has been paralleled by the creation of numerous public biological databases through which the scientific community can mine the sequences and annotations of genomes, transcriptomes, and proteomes of multiple species. However, to find the appropriate databases and bioinformatic tools for respective inquiries and aims can be challenging. Here, we present a compilation of DNA and protein databases, as well as bioinformatic tools for phylogenetic reconstruction and a wide range of studies on molecular evolution. We provide a protocol for information extraction from biological databases and simple phylogenetic reconstruction using probabilistic and distance methods, facilitating the study of biodiversity and evolution at the molecular level for the broad scientific community.

• Gregory (2008) — An introductory article on how to read phylogenetic trees, covering nodes, rooting, clades and common misinterpretations. Helpful for the tree terminology subsection. (primer; CC BY-NC 2.0 (link only))

  Understanding Evolutionary Trees. Gregory (2008). Evolution: Education and Outreach. doi:10.1007/s12052-008-0035-x

  Abstract

  Charles Darwin sketched his first evolutionary tree in 1837, and trees have remained a central metaphor in evolutionary biology up to the present. Today, phylogenetics—the science of constructing and evaluating hypotheses about historical patterns of descent in the form of evolutionary trees—has become pervasive within and increasingly outside evolutionary biology. Fostering skills in “tree thinking” is therefore a critical component of biological education. Conversely, misconceptions about evolutionary trees can be very detrimental to one’s understanding of the patterns and processes that have occurred in the history of life. This paper provides a basic introduction to evolutionary trees, including some guidelines for how and how not to read them. Ten of the most common misconceptions about evolutionary trees and their implications for understanding evolution are addressed.

D.4.2 Phylogenetic methods

• Aris-Brosou and Rodrigue (2019) — An introduction to substitution models, likelihood and Bayesian inference in molecular evolution. Complements the sections on maximum likelihood, Bayesian inference and substitution models. (book chapter; CC BY 4.0)

  A Not-So-Long Introduction to Computational Molecular Evolution. Aris-Brosou & Rodrigue (2019). Evolutionary Genomics. doi:10.1007/978-1-4939-9074-0_3

  Abstract

  In this chapter, we give a not-so-long and self-contained introduction to computational molecular evolution. In particular, we present the emergence of the use of likelihood-based methods, review the standard DNA substitution models, and introduce how model choice operates. We also present recent developments in inferring absolute divergence times and rates on a phylogeny, before showing how state-of-the-art models take inspiration from diffusion theory to link population genetics, which traditionally focuses at a taxonomic level below that of the species, and molecular evolution. Although this is not a cookbook chapter, we try and point to popular programs and implementations along the way.

• Kellis et al. (2021) — Lecture-note chapter from an MIT course that works through distance-based methods (UPGMA, neighbour joining) and character-based methods (parsimony, likelihood) with worked examples. (book chapter; CC BY-NC-SA 4.0)

  Molecular Evolution and Phylogenetics (Chapter 26 of Computational Biology: Genomes, Networks, and Evolution). Kellis et al. (2021). LibreTexts (MIT open textbook). https://bio.libretexts.org/Bookshelves/Computational_Biology/Book%3A_Computational_Biology_-_Genomes_Networks_and_Evolution_(Kellis_et_al.)/26%3A_Molecular_Evolution_and_Phylogenetics

• Reis and Yang (2019) — A tutorial on Bayesian divergence-time estimation with relaxed clocks and fossil calibrations in MCMCtree. Useful for the Bayesian inference subsection and for how trees are dated. (book chapter; CC BY 4.0)

  Bayesian Molecular Clock Dating Using Genome-Scale Datasets. dos Reis & Yang (2019). Evolutionary Genomics. doi:10.1007/978-1-4939-9074-0_10

  Abstract

  Bayesian methods for molecular clock dating of species divergences have been greatly developed during the past decade. Advantages of the methods include the use of relaxed-clock models to describe evolutionary rate variation in the branches of a phylogenetic tree and the use of flexible fossil calibration densities to describe the uncertainty in node ages. The advent of next-generation sequencing technologies has led to a flood of genome-scale datasets for organisms belonging to all domains in the tree of life. Thus, a new era has begun where dating the tree of life using genome-scale data is now within reach. In this protocol, we explain how to use the computer program MCMCTree to perform Bayesian inference of divergence times using genome-scale datasets. We use a ten-species primate phylogeny, with a molecular alignment of over three million base pairs, as an exemplar on how to carry out the analysis. We pay particular attention to how to set up the analysis and the priors and how to diagnose the MCMC algorithm used to obtain the posterior estimates of divergence times and evolutionary rates.

D.4.3 Alignments, genomes and conflicting trees

• Bravo et al. (2019) — A broad review of phylogenomic practice, including data collection, gene-tree heterogeneity, incomplete lineage sorting and coalescent-based species-tree methods. Covers most of the topics of this section. (review; CC BY 4.0)

  Embracing heterogeneity: coalescing the Tree of Life and the future of phylogenomics. Bravo et al. (2019). PeerJ. doi:10.7717/peerj.6399

  Abstract

  Building the Tree of Life (ToL) is a major challenge of modern biology, requiring advances in cyberinfrastructure, data collection, theory, and more. Here, we argue that phylogenomics stands to benefit by embracing the many heterogeneous genomic signals emerging from the first decade of large-scale phylogenetic analysis spawned by high-throughput sequencing (HTS). Such signals include those most commonly encountered in phylogenomic datasets, such as incomplete lineage sorting, but also those reticulate processes emerging with greater frequency, such as recombination and introgression. Here we focus specifically on how phylogenetic methods can accommodate the heterogeneity incurred by such population genetic processes; we do not discuss phylogenetic methods that ignore such processes, such as concatenation or supermatrix approaches or supertrees. We suggest that methods of data acquisition and the types of markers used in phylogenomics will remain restricted until a posteriori methods of marker choice are made possible with routine whole-genome sequencing of taxa of interest. We discuss limitations and potential extensions of a model supporting innovation in phylogenomics today, the multispecies coalescent model (MSC). Macroevolutionary models that use phylogenies, such as character mapping, often ignore the heterogeneity on which building phylogenies increasingly rely and suggest that assimilating such heterogeneity is an important goal moving forward. Finally, we argue that an integrative cyberinfrastructure linking all steps of the process of building the ToL, from specimen acquisition in the field to publication and tracking of phylogenomic data, as well as a culture that values contributors at each step, are essential for progress.

• Liu et al. (2019) — Explains the multispecies coalescent and the main summary and full-likelihood methods for estimating species trees from many gene trees. Useful background for the gene-tree/species-tree subsection. (book chapter; CC BY 4.0)

  Modern Phylogenomics: Building Phylogenetic Trees Using the Multispecies Coalescent Model. Liu et al. (2019). Evolutionary Genomics. doi:10.1007/978-1-4939-9074-0_7

  Abstract

  The multispecies coalescent (MSC) model provides a compelling framework for building phylogenetic trees from multilocus DNA sequence data. The pure MSC is best thought of as a special case of so-called "multispecies network coalescent" models, in which gene flow is allowed among branches of the tree, whereas MSC methods assume there is no gene flow between diverging species. Early implementations of the MSC, such as "parsimony" or "democratic vote" approaches to combining information from multiple gene trees, as well as concatenation, in which DNA sequences from multiple gene trees are combined into a single "supergene," were quickly shown to be inconsistent in some regions of tree space, in so far as they converged on the incorrect species tree as more gene trees and sequence data were accumulated. The anomaly zone, a region of tree space in which the most frequent gene tree is different from the species tree, is one such region where many so-called "coalescent" methods are inconsistent. Second-generation implementations of the MSC employed Bayesian or likelihood models; these are consistent in all regions of gene tree space, but Bayesian methods in particular are incapable of handling the large phylogenomic data sets currently available. Two-step methods, such as MP-EST and ASTRAL, in which gene trees are first estimated and then combined to estimate an overarching species tree, are currently popular in part because they can handle large phylogenomic data sets. These methods are consistent in the anomaly zone but can sometimes provide inappropriate measures of tree support or apportion error and signal in the data inappropriately. MP-EST in particular employs a likelihood model which can be conveniently manipulated to perform statistical tests of competing species trees, incorporating the likelihood of the collected gene trees on each species tree in a likelihood ratio test. Such tests provide a useful alternative to the multilocus bootstrap, which only indirectly tests the appropriateness of competing species trees. We illustrate these tests and implementations of the MSC with examples and suggest that MSC methods are a useful class of models effectively using information from multiple loci to build phylogenetic trees.

• Bjornson et al. (2024) — Reviews hybridisation, introgression and horizontal transfer as sources of gene-tree conflict, and the methods (including phylogenetic networks) used to detect them. Relevant to the subsection on phylogenetic networks. (review; CC BY 4.0)

  Reticulate evolution: Detection and utility in the phylogenomics era. Bjornson et al. (2024). Molecular Phylogenetics and Evolution. doi:10.1016/j.ympev.2024.108197

  Abstract

  Phylogenomics has enriched our understanding that the Tree of Life can have network-like or reticulate structures among some taxa and genes. Two non-vertical modes of evolution - hybridization/introgression and horizontal gene transfer - deviate from a strictly bifurcating tree model, causing non-treelike patterns. However, these reticulate processes can produce similar patterns to incomplete lineage sorting or recombination, potentially leading to ambiguity. Here, we present a brief overview of a phylogenomic workflow for inferring organismal histories and compare methods for distinguishing modes of reticulate evolution. We discuss how the timing of coalescent events can help disentangle introgression from incomplete lineage sorting and how horizontal gene transfer events can help determine the relative timing of speciation events. In doing so, we identify pitfalls of certain methods and discuss how to extend their utility across the Tree of Life. Workflows, methods, and future directions discussed herein underscore the need to embrace reticulate evolutionary patterns for understanding the timing and rates of evolutionary events, providing a clearer view of life's history.

D.4.4 The tree of life

• Spang et al. (2022) — A broad review of how genomics and phylogenomics have reshaped the tree of life, including LUCA, the root, the archaeal origin of eukaryotes and the place of viruses. (review; CC BY 4.0)

  Evolving Perspective on the Origin and Diversification of Cellular Life and the Virosphere. Spang et al. (2022). Genome Biology and Evolution. doi:10.1093/gbe/evac034

  Abstract

  The tree of life (TOL) is a powerful framework to depict the evolutionary history of cellular organisms through time, from our microbial origins to the diversification of multicellular eukaryotes that shape the visible biosphere today. During the past decades, our perception of the TOL has fundamentally changed, in part, due to profound methodological advances, which allowed a more objective approach to studying organismal and viral diversity and led to the discovery of major new branches in the TOL as well as viral lineages. Phylogenetic and comparative genomics analyses of these data have, among others, revolutionized our understanding of the deep roots and diversity of microbial life, the origin of the eukaryotic cell, eukaryotic diversity, as well as the origin, and diversification of viruses. In this review, we provide an overview of some of the recent discoveries on the evolutionary history of cellular organisms and their viruses and discuss a variety of complementary techniques that we consider crucial for making further progress in our understanding of the TOL and its interconnection with the virosphere.

• Schleper and Rodrigues-Oliveira (2026) — A recent review of the Asgard archaea and the evidence for the two-domain tree, in which eukaryotes emerged from within the archaea. Directly relevant to the subsection on two domains and the origin of eukaryotes. (review; CC BY 4.0)

  Asgard archaea: have we found our microbial ancestors?. Schleper & Rodrigues-Oliveira (2026). The EMBO Journal. doi:10.1038/s44318-026-00719-x

  Abstract

  The discovery of Asgard archaea about a decade ago has greatly reshaped our understanding of archaeal evolution and the origin of eukaryotes. Asgards are currently thought to be the closest prokaryotic relatives of eukaryotes and to represent the archaeal host lineage that participated in the endosymbiotic event leading to the first eukaryotic cell. The presence of numerous eukaryotic signature proteins in Asgard genomes supports this view and provides important insights into the deep evolutionary roots of eukaryotic cellular complexity. However, the close relationship between archaea and eukaryotes had been observed for decades, based on features that are shared in different molecular processes. This review discusses the discovery of Asgard archaea in the broader context of archaeal molecular and cellular biology and highlights how earlier findings foreshadowed their emergence. Primarily targeted at newcomers to the field, the review provides an overview of evolutionary innovations across the Archaea domain and discusses molecular and cellular features of cultivated Asgard strains in light of previous archaeal research.

• Puigbò et al. (2019) — Describes how thousands of prokaryotic gene trees can be compared to find a central tree-like signal despite horizontal gene transfer. Relevant to the subsection on whether the tree of life is a tree or a network. (book chapter; CC BY 4.0)

  Genome-Wide Comparative Analysis of Phylogenetic Trees: The Prokaryotic Forest of Life. Puigbò et al. (2019). Evolutionary Genomics. doi:10.1007/978-1-4939-9074-0_8

  Abstract

  Genome-wide comparison of phylogenetic trees is becoming an increasingly common approach in evolutionary genomics, and a variety of approaches for such comparison have been developed. In this article we present several methods for comparative analysis of large numbers of phylogenetic trees. To compare phylogenetic trees taking into account the bootstrap support for each internal branch, the boot-split distance (BSD) method is introduced as an extension of the previously developed split distance (SD) method for tree comparison. The BSD method implements the straightforward idea that comparison of phylogenetic trees can be made more robust by treating tree splits differentially depending on the bootstrap support. Approaches are also introduced for detecting treelike and netlike evolutionary trends in the phylogenetic Forest of Life (FOL), i.e., the entirety of the phylogenetic trees for conserved genes of prokaryotes. The principal method employed for this purpose includes mapping quartets of species onto trees to calculate the support of each quartet topology and so to quantify the tree and net contributions to the distances between species. We describe the applications methods used to analyze the FOL and the results obtained with these methods. These results support the concept of the Tree of Life (TOL) as a central evolutionary trend in the FOL as opposed to the traditional view of the TOL as a "species tree."

D.5 Genomics

Further explanations for Chapter 5.

D.5.1 Sequencing methods

• Heather and Chain (2016) — A historical review of DNA sequencing from Sanger and Maxam–Gilbert through second-generation platforms to single-molecule methods. Useful for placing the three generations of the section in context. (review; CC BY 4.0)

  The sequence of sequencers: The history of sequencing DNA. Heather & Chain (2016). Genomics. doi:10.1016/j.ygeno.2015.11.003

  Abstract

  Determining the order of nucleic acid residues in biological samples is an integral component of a wide variety of research applications. Over the last fifty years large numbers of researchers have applied themselves to the production of techniques and technologies to facilitate this feat, sequencing DNA and RNA molecules. This time-scale has witnessed tremendous changes, moving from sequencing short oligonucleotides to millions of bases, from struggling towards the deduction of the coding sequence of a single gene to rapid and widely available whole genome sequencing. This article traverses those years, iterating through the different generations of sequencing technology, highlighting some of the key discoveries, researchers, and sequences along the way.

• Satam et al. (2023) — A broad overview of current short- and long-read sequencing platforms, their chemistry, read lengths and error profiles, with an introduction to typical applications. (review; CC BY 4.0)

  Next-Generation Sequencing Technology: Current Trends and Advancements. Satam et al. (2023). Biology. doi:10.3390/biology12070997

  Abstract

  The advent of next-generation sequencing (NGS) has brought about a paradigm shift in genomics research, offering unparalleled capabilities for analyzing DNA and RNA molecules in a high-throughput and cost-effective manner. This transformative technology has swiftly propelled genomics advancements across diverse domains. NGS allows for the rapid sequencing of millions of DNA fragments simultaneously, providing comprehensive insights into genome structure, genetic variations, gene expression profiles, and epigenetic modifications. The versatility of NGS platforms has expanded the scope of genomics research, facilitating studies on rare genetic diseases, cancer genomics, microbiome analysis, infectious diseases, and population genetics. Moreover, NGS has enabled the development of targeted therapies, precision medicine approaches, and improved diagnostic methods. This review provides an insightful overview of the current trends and recent advancements in NGS technology, highlighting its potential impact on diverse areas of genomic research. Moreover, the review delves into the challenges encountered and future directions of NGS technology, including endeavors to enhance the accuracy and sensitivity of sequencing data, the development of novel algorithms for data analysis, and the pursuit of more efficient, scalable, and cost-effective solutions that lie ahead.

• Amarasinghe et al. (2020) — Reviews PacBio and Oxford Nanopore technologies and the computational steps that follow, such as base calling, error correction and alignment of long reads. Complements the section on third-generation sequencing. (review; CC BY 4.0)

  Opportunities and challenges in long-read sequencing data analysis. Amarasinghe et al. (2020). Genome Biology. doi:10.1186/s13059-020-1935-5

  Abstract

  Long-read technologies are overcoming early limitations in accuracy and throughput, broadening their application domains in genomics. Dedicated analysis tools that take into account the characteristics of long-read data are thus required, but the fast pace of development of such tools can be overwhelming. To assist in the design and analysis of long-read sequencing projects, we review the current landscape of available tools and present an online interactive database, long-read-tools.org, to facilitate their browsing. We further focus on the principles of error correction, base modification detection, and long-read transcriptomics analysis and highlight the challenges that remain.

D.5.2 Genome assembly

• Jung et al. (2020) — A short, practical guide to assembling and annotating a genome, from choosing a sequencing strategy to quality assessment. Gives a hands-on view of the workflow described in this section and in the gene prediction section. (tutorial; CC BY 4.0)

  Twelve quick steps for genome assembly and annotation in the classroom. Jung et al. (2020). PLOS Computational Biology. doi:10.1371/journal.pcbi.1008325

  Abstract

  Eukaryotic genome sequencing and de novo assembly, once the exclusive domain of well-funded international consortia, have become increasingly affordable, thus fitting the budgets of individual research groups. Third-generation long-read DNA sequencing technologies are increasingly used, providing extensive genomic toolkits that were once reserved for a few select model organisms. Generating high-quality genome assemblies and annotations for many aquatic species still presents significant challenges due to their large genome sizes, complexity, and high chromosome numbers. Indeed, selecting the most appropriate sequencing and software platforms and annotation pipelines for a new genome project can be daunting because tools often only work in limited contexts. In genomics, generating a high-quality genome assembly/annotation has become an indispensable tool for better understanding the biology of any species. Herein, we state 12 steps to help researchers get started in genome projects by presenting guidelines that are broadly applicable (to any species), sustainable over time, and cover all aspects of genome assembly and annotation projects from start to finish. We review some commonly used approaches, including practical methods to extract high-quality DNA and choices for the best sequencing platforms and library preparations. In addition, we discuss the range of potential bioinformatics pipelines, including structural and functional annotations (e.g., transposable elements and repetitive sequences). This paper also includes information on how to build a wide community for a genome project, the importance of data management, and how to make the data and results Findable, Accessible, Interoperable, and Reusable (FAIR) by submitting them to a public repository and sharing them with the research community.

• Dominguez Del Angel et al. (2018) — An introductory guide covering planning, data types, assembly, quality assessment and annotation of eukaryotic genomes. Useful for understanding the practical choices behind assembly projects. (tutorial; CC BY 4.0)

  Ten steps to get started in Genome Assembly and Annotation. Dominguez Del Angel et al. (2018). F1000Research. doi:10.12688/f1000research.13598.1

  Abstract

  As a part of the ELIXIR-EXCELERATE efforts in capacity building, we present here 10 steps to facilitate researchers getting started in genome assembly and genome annotation. The guidelines given are broadly applicable, intended to be stable over time, and cover all aspects from start to finish of a general assembly and annotation project. Intrinsic properties of genomes are discussed, as is the importance of using high quality DNA. Different sequencing technologies and generally applicable workflows for genome assembly are also detailed. We cover structural and functional annotation and encourage readers to also annotate transposable elements, something that is often omitted from annotation workflows. The importance of data management is stressed, and we give advice on where to submit data and how to make your results Findable, Accessible, Interoperable, and Reusable (FAIR).

• Ekblom and Wolf (2014) — A readable guide for biologists to sequencing strategies, de Bruijn graph and overlap assembly, scaffolding, assembly quality metrics and annotation. Some technology details are dated, but the concepts match this chapter well. (review; CC BY 3.0)

  A field guide to whole-genome sequencing, assembly and annotation. Ekblom & Wolf (2014). Evolutionary Applications. doi:10.1111/eva.12178

  Abstract

  Genome sequencing projects were long confined to biomedical model organisms and required the concerted effort of large consortia. Rapid progress in high-throughput sequencing technology and the simultaneous development of bioinformatic tools have democratized the field. It is now within reach for individual research groups in the eco-evolutionary and conservation community to generate de novo draft genome sequences for any organism of choice. Because of the cost and considerable effort involved in such an endeavour, the important first step is to thoroughly consider whether a genome sequence is necessary for addressing the biological question at hand. Once this decision is taken, a genome project requires careful planning with respect to the organism involved and the intended quality of the genome draft. Here, we briefly review the state of the art within this field and provide a step-by-step introduction to the workflow involved in genome sequencing, assembly and annotation with particular reference to large and complex genomes. This tutorial is targeted at scientists with a background in conservation genetics, but more generally, provides useful practical guidance for researchers engaging in whole-genome sequencing projects.

D.5.3 Mapping and variant calling

• Alser et al. (2021) — Surveys read-mapping algorithms, including indexing with hash tables, the Burrows–Wheeler transform and minimizers, seeding, filtering and alignment, and how they have followed sequencing technology. Explains the algorithms behind the read-mapping subsection. (review; CC BY 4.0)

  Technology dictates algorithms: recent developments in read alignment. Alser et al. (2021). Genome Biology. doi:10.1186/s13059-021-02443-7

  Abstract

  Aligning sequencing reads onto a reference is an essential step of the majority of genomic analysis pipelines. Computational algorithms for read alignment have evolved in accordance with technological advances, leading to today's diverse array of alignment methods. We provide a systematic survey of algorithmic foundations and methodologies across 107 alignment methods, for both short and long reads. We provide a rigorous experimental evaluation of 11 read aligners to demonstrate the effect of these underlying algorithms on speed and efficiency of read alignment. We discuss how general alignment algorithms have been tailored to the specific needs of various domains in biology.

• Zverinova and Guryev (2021) — Reviews the steps from aligned reads to called variants, covering small variants and structural variants, sources of error and benchmarking. A good companion to the variant-calling subsection. (review; CC BY 4.0)

  Variant calling: Considerations, practices, and developments. Zverinova & Guryev (2021). Human Mutation. doi:10.1002/humu.24311

  Abstract

  The success of many clinical, association, or population genetics studies critically relies on properly performed variant calling step. The variety of modern genomics protocols, techniques, and platforms makes our choices of methods and algorithms difficult and there is no "one size fits all" solution for study design and data analysis. In this review, we discuss considerations that need to be taken into account while designing the study and preparing for the experiments. We outline the variety of variant types that can be detected using sequencing approaches and highlight some specific requirements and basic principles of their detection. Finally, we cover interesting developments that enable variant calling for a broad range of applications in the genomics field. We conclude by discussing technological and algorithmic advances that have the potential to change the ways of calling DNA variants in the nearest future.

• Koboldt (2020) — Describes a practical variant-calling workflow (alignment, duplicate marking, base quality recalibration, calling and filtering) for germline and somatic variants. Clinically oriented, but the workflow is general. (review; CC BY 4.0)

  Best practices for variant calling in clinical sequencing. Koboldt (2020). Genome Medicine. doi:10.1186/s13073-020-00791-w

  Abstract

  Next-generation sequencing technologies have enabled a dramatic expansion of clinical genetic testing both for inherited conditions and diseases such as cancer. Accurate variant calling in NGS data is a critical step upon which virtually all downstream analysis and interpretation processes rely. Just as NGS technologies have evolved considerably over the past 10 years, so too have the software tools and approaches for detecting sequence variants in clinical samples. In this review, I discuss the current best practices for variant calling in clinical sequencing studies, with a particular emphasis on trio sequencing for inherited disorders and somatic mutation detection in cancer patients. I describe the relative strengths and weaknesses of panel, exome, and whole-genome sequencing for variant detection. Recommended tools and strategies for calling variants of different classes are also provided, along with guidance on variant review, validation, and benchmarking to ensure optimal performance. Although NGS technologies are continually evolving, and new capabilities (such as long-read single-molecule sequencing) are emerging, the "best practice" principles in this review should be relevant to clinical variant calling in the long term.

D.5.4 Gene prediction

• Ejigu and Jung (2020) — Reviews structural and functional genome annotation, including ab initio gene finders, evidence-based and combined pipelines, for prokaryotes and eukaryotes. Matches the scope of this section. (review; CC BY 4.0)

  Review on the Computational Genome Annotation of Sequences Obtained by Next-Generation Sequencing. Ejigu & Jung (2020). Biology. doi:10.3390/biology9090295

  Abstract

  Next-Generation Sequencing (NGS) has made it easier to obtain genome-wide sequence data and it has shifted the research focus into genome annotation. The challenging tasks involved in annotation rely on the currently available tools and techniques to decode the information contained in nucleotide sequences. This information will improve our understanding of general aspects of life and evolution and improve our ability to diagnose genetic disorders. Here, we present a summary of both structural and functional annotations, as well as the associated comparative annotation tools and pipelines. We highlight visualization tools that immensely aid the annotation process and the contributions of the scientific community to the annotation. Further, we discuss quality-control practices and the need for re-annotation, and highlight the future of annotation.

• Salzberg (2019) — A short commentary on why gene annotation of new genomes remains error-prone, for example through contamination and incomplete evidence. Useful for discussing the limits of automated annotation. (perspective; CC BY 4.0)

  Next-generation genome annotation: we still struggle to get it right. Salzberg (2019). Genome Biology. doi:10.1186/s13059-019-1715-2

  Abstract

  While the genome sequencing revolution has led to the sequencing and assembly of many thousands of new genomes, genome annotation still uses very nearly the same technology that we have used for the past two decades. The sheer number of genomes necessitates the use of fully automated procedures for annotation, but errors in annotation are just as prevalent as they were in the past, if not more so. How are we to solve this growing problem?

D.5.5 Metagenomics

• Wooley et al. (2010) — An introduction to metagenomics for computational biologists, covering sampling, 16S studies, shotgun sequencing, assembly, binning and functional analysis. Older, but explains the basic concepts clearly. (primer; CC BY 4.0)

  A Primer on Metagenomics. Wooley et al. (2010). PLoS Computational Biology. doi:10.1371/journal.pcbi.1000667

  Abstract

  Metagenomics is a discipline that enables the genomic study of uncultured microorganisms. Faster, cheaper sequencing technologies and the ability to sequence uncultured microbes sampled directly from their habitats are expanding and transforming our view of the microbial world. Distilling meaningful information from the millions of new genomic sequences presents a serious challenge to bioinformaticians. In cultured microbes, the genomic data come from a single clone, making sequence assembly and annotation tractable. In metagenomics, the data come from heterogeneous microbial communities, sometimes containing more than 10,000 species, with the sequence data being noisy and partial. From sampling, to assembly, to gene calling and function prediction, bioinformatics faces new demands in interpreting voluminous, noisy, and often partial sequence data. Although metagenomics is a relative newcomer to science, the past few years have seen an explosion in computational methods applied to metagenomic-based research. It is therefore not within the scope of this article to provide an exhaustive review. Rather, we provide here a concise yet comprehensive introduction to the current computational requirements presented by metagenomics, and review the recent progress made. We also note whether there is software that implements any of the methods presented here, and briefly review its utility. Nevertheless, it would be useful if readers of this article would avail themselves of the comment section provided by this journal, and relate their own experiences. Finally, the last section of this article provides a few representative studies illustrating different facets of recent scientific discoveries made using metagenomics.

• Pérez-Cobas et al. (2020) — Compares amplicon (marker gene) and shotgun metagenomics and describes the analysis steps and tools for each. Covers the amplicon and shotgun subsections. (review; CC BY 4.0)

  Metagenomic approaches in microbial ecology: an update on whole-genome and marker gene sequencing analyses. Pérez-Cobas et al. (2020). Microbial Genomics. doi:10.1099/mgen.0.000409

  Abstract

  Metagenomics and marker gene approaches, coupled with high-throughput sequencing technologies, have revolutionized the field of microbial ecology. Metagenomics is a culture-independent method that allows the identification and characterization of organisms from all kinds of samples. Whole-genome shotgun sequencing analyses the total DNA of a chosen sample to determine the presence of micro-organisms from all domains of life and their genomic content. Importantly, the whole-genome shotgun sequencing approach reveals the genomic diversity present, but can also give insights into the functional potential of the micro-organisms identified. The marker gene approach is based on the sequencing of a specific gene region. It allows one to describe the microbial composition based on the taxonomic groups present in the sample. It is frequently used to analyse the biodiversity of microbial ecosystems. Despite its importance, the analysis of metagenomic sequencing and marker gene data is quite a challenge. Here we review the primary workflows and software used for both approaches and discuss the current challenges in the field.

• Lapidus and Korobeynikov (2021) — Reviews why metagenome assembly is harder than single-genome assembly and how metagenomic assemblers and binning lead to metagenome-assembled genomes. (review; CC BY 4.0)

  Metagenomic Data Assembly – The Way of Decoding Unknown Microorganisms. Lapidus & Korobeynikov (2021). Frontiers in Microbiology. doi:10.3389/fmicb.2021.613791

  Abstract

  Metagenomics is a segment of conventional microbial genomics dedicated to the sequencing and analysis of combined genomic DNA of entire environmental samples. The most critical step of the metagenomic data analysis is the reconstruction of individual genes and genomes of the microorganisms in the communities using metagenomic assemblers - computational programs that put together small fragments of sequenced DNA generated by sequencing instruments. Here, we describe the challenges of metagenomic assembly, a wide spectrum of applications in which metagenomic assemblies were used to better understand the ecology and evolution of microbial ecosystems, and present one of the most efficient microbial assemblers, SPAdes that was upgraded to become applicable for metagenomics.

D.5.6 Regulatory and epigenomic sequencing data

• Ma and Zhang (2020) — An open review of how ChIP-seq and ATAC-seq work and how their data are analysed, from peak calling to integration with other data. (review; CC BY 4.0)

  Profiling chromatin regulatory landscape: insights into the development of ChIP-seq and ATAC-seq. Ma & Zhang (2020). Molecular Biomedicine. doi:10.1186/s43556-020-00009-w

  Abstract

  Chromatin regulatory landscape plays a critical role in many disease processes and embryo development. Epigenome sequencing technologies such as chromatin immunoprecipitation sequencing (ChIP-seq) and assay for transposase-accessible chromatin with high-throughput sequencing (ATAC-seq) have enabled us to dissect the pan-genomic regulatory landscape of cells and tissues in both time and space dimensions by detecting specific chromatin state and its corresponding transcription factors. Pioneered by the advancement of chromatin immunoprecipitation-chip (ChIP-chip) technology, abundant epigenome profiling technologies have become available such as ChIP-seq, DNase I hypersensitive site sequencing (DNase-seq), ATAC-seq and so on. The advent of single-cell sequencing has revolutionized the next-generation sequencing, applications in single-cell epigenetics are enriched rapidly. Epigenome sequencing technologies have evolved from low-throughput to high-throughput and from bulk sample to the single-cell scope, which unprecedentedly benefits scientists to interpret life from different angles. In this review, after briefly introducing the background knowledge of epigenome biology, we discuss the development of epigenome sequencing technologies, especially ChIP-seq & ATAC-seq and their current applications in scientific research. Finally, we provide insights into future applications and challenges.

D.5.7 Human genetics: finding disease genes and variants

• Visscher et al. (2017) — A balanced account of what genome-wide association studies have and have not achieved, including heritability, polygenicity and translation. (review; free to read (Elsevier open archive))

  10 Years of GWAS Discovery: Biology, Function, and Translation. Visscher et al. (2017). The American Journal of Human Genetics. doi:10.1016/j.ajhg.2017.06.005

  Abstract

  Application of the experimental design of genome-wide association studies (GWASs) is now 10 years old (young), and here we review the remarkable range of discoveries it has facilitated in population and complex-trait genetics, the biology of diseases, and translation toward new therapeutics. We predict the likely discoveries in the next 10 years, when GWASs will be based on millions of samples with array data imputed to a large fully sequenced reference panel and on hundreds of thousands of samples with whole-genome sequencing data.

D.6 Genomes

Further explanations for Chapter 6.

D.6.1 Background: genes, genomes and cells

• Clark, Douglas, Choi (OpenStax) (2018a) — Open textbook chapter on the double helix, base pairing, replication and DNA repair, at introductory undergraduate level. Useful for revising the nucleic-acid basics assumed by the rest of the chapter. (book chapter; CC BY-NC-SA 4.0)

  Chapter 14: DNA Structure and Function (Biology 2e). Clark, Douglas, Choi (OpenStax) (2018). OpenStax Biology 2e. https://openstax.org/books/biology-2e/pages/14-introduction

• Clark, Douglas, Choi (OpenStax) (2018b) — Open textbook chapter on the genetic code, transcription, RNA processing and translation in prokaryotes and eukaryotes. Covers the central dogma at the level summarised in this section. (book chapter; CC BY-NC-SA 4.0)

  Chapter 15: Genes and Proteins (Biology 2e). Clark, Douglas, Choi (OpenStax) (2018). OpenStax Biology 2e. https://openstax.org/books/biology-2e/pages/15-introduction

• Heather and Chain (2016) — A short historical review of DNA sequencing, from Sanger and Maxam-Gilbert sequencing to second- and third-generation methods. Gives background for the experimental methods mentioned in this section. (review; CC BY 4.0)

  The sequence of sequencers: The history of sequencing DNA. Heather & Chain (2016). Genomics. doi:10.1016/j.ygeno.2015.11.003

  Abstract

  Determining the order of nucleic acid residues in biological samples is an integral component of a wide variety of research applications. Over the last fifty years large numbers of researchers have applied themselves to the production of techniques and technologies to facilitate this feat, sequencing DNA and RNA molecules. This time-scale has witnessed tremendous changes, moving from sequencing short oligonucleotides to millions of bases, from struggling towards the deduction of the coding sequence of a single gene to rapid and widely available whole genome sequencing. This article traverses those years, iterating through the different generations of sequencing technology, highlighting some of the key discoveries, researchers, and sequences along the way.

D.6.2 From the first genome projects

• Taylor et al. (2024) — Reviews how the incomplete reference from the Human Genome Project was completed by the Telomere-to-Telomere assembly and extended by the Human Pangenome Reference. Connects the genome projects in this section with the pangenome discussion in the next. (review; CC BY 4.0)

  Beyond the Human Genome Project: The Age of Complete Human Genome Sequences and Pangenome References. Taylor et al. (2024). Annual Review of Genomics and Human Genetics. doi:10.1146/annurev-genom-021623-081639

  Abstract

  The Human Genome Project was an enormous accomplishment, providing a foundation for countless explorations into the genetics and genomics of the human species. Yet for many years, the human genome reference sequence remained incomplete and lacked representation of human genetic diversity. Recently, two major advances have emerged to address these shortcomings: complete gap-free human genome sequences, such as the one developed by the Telomere-to-Telomere Consortium, and high-quality pangenomes, such as the one developed by the Human Pangenome Reference Consortium. Facilitated by advances in long-read DNA sequencing and genome assembly algorithms, complete human genome sequences resolve regions that have been historically difficult to sequence, including centromeres, telomeres, and segmental duplications. In parallel, pangenomes capture the extensive genetic diversity across populations worldwide. Together, these advances usher in a new era of genomics research, enhancing the accuracy of genomic analysis, paving the path for precision medicine, and contributing to deeper insights into human biology.

• Birney (2021) — A brief perspective on why the Human Genome Project was organised internationally and kept its data freely available. Useful context for the Bermuda Principles and open data in genomics. (perspective; CC BY 4.0)

  The International Human Genome Project. Birney (2021). Human Molecular Genetics. doi:10.1093/hmg/ddab198

  Abstract

  The human genome project was conceived and executed as an international project, due to both pragmatic and principled reasons. This internationality has served the project well, with the resulting human genome being freely available for all researchers in all countries. Over time the reference human genome will likely have to evolve to a graph genome, and tap into more diverse sequences worldwide. A similar international mindset underpins data analysis for the interpretation of the human genome from basic to clinical research.

• Pellicer et al. (2018) — Reviews the range of genome sizes in eukaryotes and especially land plants, and the roles of polyploidy and transposable elements in genome size change. Gives concrete examples for the C-value paradox. (review; CC BY 4.0)

  Genome Size Diversity and Its Impact on the Evolution of Land Plants. Pellicer et al. (2018). Genes. doi:10.3390/genes9020088

  Abstract

  Genome size is a biodiversity trait that shows staggering diversity across eukaryotes, varying over 64,000-fold. Of all major taxonomic groups, land plants stand out due to their staggering genome size diversity, ranging ca. 2400-fold. As our understanding of the implications and significance of this remarkable genome size diversity in land plants grows, it is becoming increasingly evident that this trait plays not only an important role in shaping the evolution of plant genomes, but also in influencing plant community assemblages at the ecosystem level. Recent advances and improvements in novel sequencing technologies, as well as analytical tools, make it possible to gain critical insights into the genomic and epigenetic mechanisms underpinning genome size changes. In this review we provide an overview of our current understanding of genome size diversity across the different land plant groups, its implications on the biology of the genome and what future directions need to be addressed to fill key knowledge gaps.

D.6.3 Differences across the tree of life

• Youssef et al. (2019) — Open-access book chapter introducing genome structure and content across the tree of life, including how prokaryotic and eukaryotic genomes differ. A broad starting point for this section. (book chapter; CC BY 4.0)

  Introduction to Genome Biology and Diversity. Youssef et al. (2019). Evolutionary Genomics. doi:10.1007/978-1-4939-9074-0_1

  Abstract

  Organisms display astonishing levels of cell and molecular diversity, including genome size, shape, and architecture. In this chapter, we review how the genome can be viewed as both a structural and an informational unit of biological diversity and explicitly define our intended meaning of genetic information. A brief overview of the characteristic features of bacterial, archaeal, and eukaryotic cell types and viruses sets the stage for a review of the differences in organization, size, and packaging strategies of their genomes. We include a detailed review of genetic elements found outside the primary chromosomal structures, as these provide insights into how genomes are sometimes viewed as incomplete informational entities. Lastly, we reassess the definition of the genome in light of recent advancements in our understanding of the diversity of genomic structures and the mechanisms by which genetic information is expressed within the cell. Collectively, these topics comprise a good introduction to genome biology for the newcomer to the field and provide a valuable reference for those developing new statistical or computation methods in genomics. This review also prepares the reader for anticipated transformations in thinking as the field of genome biology progresses.

• Bobay and Ochman (2017) — Short review of why bacterial genomes are compact and gene-dense while eukaryotic genomes accumulate non-functional sequence, explained by a deletion bias in bacteria and by genetic drift. Complements the subsection on prokaryotic genome organisation and the population-genetic arguments in the next section. (review; CC BY 4.0)

  The Evolution of Bacterial Genome Architecture. Bobay & Ochman (2017). Frontiers in Genetics. doi:10.3389/fgene.2017.00072

  Abstract

  The genome architecture of bacteria and eukaryotes evolves in opposite directions when subject to genetic drift, a difference that can be ascribed to the fact that bacteria exhibit a mutational bias that deletes superfluous sequences, whereas eukaryotes are biased toward large insertions. Expansion of eukaryotic genomes occurs through the addition of non-functional sequences, such as repetitive sequences and transposable elements, whereas variation in bacterial genome size is largely due to the acquisition and loss of functional accessory genes. These properties create the situation in which eukaryotes with very similar numbers of genes can have vastly different genome sizes, while in bacteria, gene number scales linearly with genome size. Some bacterial genomes, however, particularly those of species that undergo bottlenecks due to recent association with hosts, accumulate pseudogenes and mobile elements, conferring them a low gene content relative to their genome size. These non-functional sequences are gradually eroded and eliminated after long-term association with hosts, with the result that obligate symbionts have the smallest genomes of any cellular organism. The architecture of bacterial genomes is shaped by complex and diverse processes, but for most bacterial species, genome size is governed by a non-adaptive process, i.e., genetic drift coupled with a mutational bias toward deletions. Thus, bacteria with small effective population sizes typically have the smallest genomes. Some marine bacteria counter this near-universal trend: despite having immense population sizes, selection, not drift, acts to reduce genome size in response to metabolic constraints in their nutrient-limited environment.

• Koonin (2015) — Discusses the origin of the eukaryotic cell from an archaeal host and a bacterial endosymbiont, from a comparative-genomics perspective. Relevant to the subsection on organelle genomes and endosymbiosis. (review; CC BY 4.0)

  Origin of eukaryotes from within archaea, archaeal eukaryome and bursts of gene gain: eukaryogenesis just made easier?. Koonin (2015). Philosophical Transactions of the Royal Society B: Biological Sciences. doi:10.1098/rstb.2014.0333

  Abstract

  The origin of eukaryotes is a fundamental, forbidding evolutionary puzzle. Comparative genomic analysis clearly shows that the last eukaryotic common ancestor (LECA) possessed most of the signature complex features of modern eukaryotic cells, in particular the mitochondria, the endomembrane system including the nucleus, an advanced cytoskeleton and the ubiquitin network. Numerous duplications of ancestral genes, e.g. DNA polymerases, RNA polymerases and proteasome subunits, also can be traced back to the LECA. Thus, the LECA was not a primitive organism and its emergence must have resulted from extensive evolution towards cellular complexity. However, the scenario of eukaryogenesis, and in particular the relationship between endosymbiosis and the origin of eukaryotes, is far from being clear. Four recent developments provide new clues to the likely routes of eukaryogenesis. First, evolutionary reconstructions suggest complex ancestors for most of the major groups of archaea, with the subsequent evolution dominated by gene loss. Second, homologues of signature eukaryotic proteins, such as actin and tubulin that form the core of the cytoskeleton or the ubiquitin system, have been detected in diverse archaea. The discovery of this 'dispersed eukaryome' implies that the archaeal ancestor of eukaryotes was a complex cell that might have been capable of a primitive form of phagocytosis and thus conducive to endosymbiont capture. Third, phylogenomic analyses converge on the origin of most eukaryotic genes of archaeal descent from within the archaeal evolutionary tree, specifically, the TACK superphylum. Fourth, evidence has been presented that the origin of the major archaeal phyla involved massive acquisition of bacterial genes. Taken together, these findings make the symbiogenetic scenario for the origin of eukaryotes considerably more plausible and the origin of the organizational complexity of eukaryotic cells more readily explainable than they appeared until recently.

D.6.4 Genomes in populations

• Medini et al. (2020) — Opening chapter of an open-access book, written by the authors who introduced the pan-genome concept, describing how core and accessory genomes were defined from multiple bacterial genomes. Other chapters of the same book cover pangenome ecology and selection. (book chapter; CC BY 4.0)

  The Pangenome: A Data-Driven Discovery in Biology. Medini et al. (2020). The Pangenome. doi:10.1007/978-3-030-38281-0_1

  Abstract

  An early example of Big data in biology: how a mathematical model, developed to address a practical question in vaccinology, transformed established concepts, opening biology to the “unbounded.”

• Wolf and Akey (2018) — Short review of what genome sequences have shown about Neanderthal and Denisovan admixture into modern humans, and which questions remain open. (review; CC BY 4.0)

  Outstanding questions in the study of archaic hominin admixture. Wolf & Akey (2018). PLOS Genetics. doi:10.1371/journal.pgen.1007349

  Abstract

  The complete sequencing of archaic and modern human genomes has revolutionized the study of human history and evolution. The application of paleogenomics has answered questions that were beyond the scope of archaeology alone-definitively proving admixture between archaic and modern humans. Despite the remarkable progress made in the study of archaic-modern human admixture, many outstanding questions remain. Here, we review some of these questions, which include how frequent archaic-modern human admixture was in history, to what degree drift and selection are responsible for the loss and retention of introgressed sequences in modern human genomes, and how surviving archaic sequences affect human phenotypes.

• Ahlquist et al. (2021) — Reviews a decade of work on archaic introgression, including demographic history and the methods used to detect introgressed segments. Gives more methodological detail than the previous item. (review; CC BY 4.0)

  Our Tangled Family Tree: New Genomic Methods Offer Insight into the Legacy of Archaic Admixture. Ahlquist et al. (2021). Genome Biology and Evolution. doi:10.1093/gbe/evab115

  Abstract

  The archaic ancestry present in the human genome has captured the imagination of both scientists and the wider public in recent years. This excitement is the result of new studies pushing the envelope of what we can learn from the archaic genetic information that has survived for over 50,000 years in the human genome. Here, we review the most recent ten years of literature on the topic of archaic introgression, including the current state of knowledge on Neanderthal and Denisovan introgression, as well as introgression from other as-yet unidentified archaic populations. We focus this review on four topics: 1) a reimagining of human demographic history, including evidence for multiple admixture events between modern humans, Neanderthals, Denisovans, and other archaic populations; 2) state-of-the-art methods for detecting archaic ancestry in population-level genomic data; 3) how these novel methods can detect archaic introgression in modern African populations; and 4) the functional consequences of archaic gene variants, including how those variants were co-opted into novel function in modern human populations. The goal of this review is to provide a simple-to-access reference for the relevant methods and novel data, which has changed our understanding of the relationship between our species and its siblings. This body of literature reveals the large degree to which the genetic legacy of these extinct hominins has been integrated into the human populations of today.

D.6.5 Repeats and the 3D genome

• Makałowski et al. (2019) — Open-access chapter describing the classes of transposable elements and the computational methods used to identify and annotate them. Useful for the subsection on transposable elements. (book chapter; CC BY 4.0)

  Transposable Elements: Classification, Identification, and Their Use As a Tool For Comparative Genomics. Makałowski et al. (2019). Evolutionary Genomics. doi:10.1007/978-1-4939-9074-0_6

  Abstract

  Most genomes are populated by hundreds of thousands of sequences originated from mobile elements. On the one hand, these sequences present a real challenge in the process of genome analysis and annotation. On the other hand, they are very interesting biological subjects involved in many cellular processes. Here we present an overview of transposable elements biodiversity, and we discuss different approaches to transposable elements detection and analyses.

• Garrido-Ramos (2017) — Reviews the structure, organisation, evolution and possible functions of satellite DNA, including how sequencing data are used to study it. Complements the subsection on tandem repeats. (review; CC BY 4.0)

  Satellite DNA: An Evolving Topic. Garrido-Ramos (2017). Genes. doi:10.3390/genes8090230

  Abstract

  Satellite DNA represents one of the most fascinating parts of the repetitive fraction of the eukaryotic genome. Since the discovery of highly repetitive tandem DNA in the 1960s, a lot of literature has extensively covered various topics related to the structure, organization, function, and evolution of such sequences. Today, with the advent of genomic tools, the study of satellite DNA has regained a great interest. Thus, Next-Generation Sequencing (NGS), together with high-throughput in silico analysis of the information contained in NGS reads, has revolutionized the analysis of the repetitive fraction of the eukaryotic genomes. The whole of the historical and current approaches to the topic gives us a broad view of the function and evolution of satellite DNA and its role in chromosomal evolution. Currently, we have extensive information on the molecular, chromosomal, biological, and population factors that affect the evolutionary fate of satellite DNA, knowledge that gives rise to a series of hypotheses that get on well with each other about the origin, spreading, and evolution of satellite DNA. In this paper, I review these hypotheses from a methodological, conceptual, and historical perspective and frame them in the context of chromosomal organization and evolution.

• Pal et al. (2018) — Explains how Hi-C data are generated and analysed, from normalisation to the calling of compartments, TADs and loops. Useful for understanding the methods behind 3D genome organisation. (review; CC BY 4.0)

  Hi-C analysis: from data generation to integration. Pal et al. (2018). Biophysical Reviews. doi:10.1007/s12551-018-0489-1

  Abstract

  In the epigenetics field, large-scale functional genomics datasets of ever-increasing size and complexity have been produced using experimental techniques based on high-throughput sequencing. In particular, the study of the 3D organization of chromatin has raised increasing interest, thanks to the development of advanced experimental techniques. In this context, Hi-C has been widely adopted as a high-throughput method to measure pairwise contacts between virtually any pair of genomic loci, thus yielding unprecedented challenges for analyzing and handling the resulting complex datasets. In this review, we focus on the increasing complexity of available Hi-C datasets, which parallels the adoption of novel protocol variants. We also review the complexity of the multiple data analysis steps required to preprocess Hi-C sequencing reads and extract biologically meaningful information. Finally, we discuss solutions for handling and visualizing such large genomics datasets.

D.6.6 RNA genes and regulatory elements

• Palazzo and Gregory (2014) — A readable argument, based on genome size variation, transposable elements and population genetics, for why much of the human genome is likely to be non-functional. Directly relevant to the debate after ENCODE discussed in this section. (review; CC BY 4.0)

  The Case for Junk DNA. Palazzo & Gregory (2014). PLoS Genetics. doi:10.1371/journal.pgen.1004351

  Abstract

  With the advent of deep sequencing technologies and the ability to analyze whole genome sequences and transcriptomes, there has been a growing interest in exploring putative functions of the very large fraction of the genome that is commonly referred to as ''junk DNA.''Whereas this is an issue of considerable importance in genome biology, there is an unfortunate tendency for researchers and science writers to proclaim the demise of junk DNA on a regular basis without properly addressing some of the fundamental issues that first led to the rise of the concept.In this review, we provide an overview of the major arguments that have been presented in support of the notion that a large portion of most eukaryotic genomes lacks an organism-level function.Some of these are based on observations or basic genetic principles that are decades old, whereas others stem from new knowledge regarding molecular processes such as transcription and gene regulation.

• Ponting and Haerty (2022) — A critical review of the evidence for lncRNA function, discussing the pitfalls of experimental and computational methods. Argues that most lncRNA transcripts are likely to be noise or to have minor regulatory roles. (review; CC BY 4.0)

  Genome-Wide Analysis of Human Long Noncoding RNAs: A Provocative Review. Ponting & Haerty (2022). Annual Review of Genomics and Human Genetics. doi:10.1146/annurev-genom-112921-123710

  Abstract

  Do long noncoding RNAs (lncRNAs) contribute little or substantively to human biology? To address how lncRNA loci and their transcripts, structures, interactions, and functions contribute to human traits and disease, we adopt a genome-wide perspective. We intend to provoke alternative interpretation of questionable evidence and thorough inquiry into unsubstantiated claims. We discuss pitfalls of lncRNA experimental and computational methods as well as opposing interpretations of their results. The majority of evidence, we argue, indicates that most lncRNA transcript models reflect transcriptional noise or provide minor regulatory roles, leaving relatively few human lncRNAs that contribute centrally to human development, physiology, or behavior. These important few tend to be spliced and better conserved but lack a simple syntax relating sequence to structure and mechanism, and so resist simple categorization. This genome-wide view should help investigators prioritize individual lncRNAs based on their likely contribution to human biology.

• Marchese et al. (2017) — Reviews the proposed mechanisms by which lncRNAs act, through RNA sequence and structure, protein interactions and the act of transcription itself. A useful counterpoint to the more sceptical review above. (review; CC BY 4.0)

  The multidimensional mechanisms of long noncoding RNA function. Marchese et al. (2017). Genome Biology. doi:10.1186/s13059-017-1348-2

  Abstract

  A major shift in our understanding of genome regulation has emerged recently. It is now apparent that the majority of cellular transcripts do not code for proteins, and many of them are long noncoding RNAs (lncRNAs). Increasingly, studies suggest that lncRNAs regulate gene expression through diverse mechanisms. We review emerging mechanistic views of lncRNAs in gene regulation in the cell nucleus. We discuss the functional interactions that lncRNAs establish with other molecules as well as the relationship between lncRNA transcription and function. While some of these mechanisms are specific to lncRNAs, others might be shared with other types of genes.

D.6.7 Gene order, synteny and genome rearrangements

• Liu et al. (2018) — Compares synteny-detection tools on real and simulated assemblies, and shows how assembly fragmentation and parameter choices change the inferred synteny blocks; a practical complement to the section. (primary study; CC BY 4.0)

  Inferring synteny between genome assemblies: a systematic evaluation. Liu et al. (2018). BMC Bioinformatics. doi:10.1186/s12859-018-2026-4

  Abstract

  Background Genome assemblies across all domains of life are being produced routinely. Initial analysis of a new genome usually includes annotation and comparative genomics. Synteny provides a framework in which conservation of homologous genes and gene order is identified between genomes of different species. The availability of human and mouse genomes paved the way for algorithm development in large-scale synteny mapping, which eventually became an integral part of comparative genomics. Synteny analysis is regularly performed on assembled sequences that are fragmented, neglecting the fact that most methods were developed using complete genomes. It is unknown to what extent draft assemblies lead to errors in such analysis. Results We fragmented genome assemblies of model nematodes to various extents and conducted synteny identification and downstream analysis. We first show that synteny between species can be underestimated up to 40% and find disagreements between popular tools that infer synteny blocks. This inconsistency and further demonstration of erroneous gene ontology enrichment tests raise questions about the robustness of previous synteny analysis when gold standard genome sequences remain limited. In addition, assembly scaffolding using a reference guided approach with a closely related species may result in chimeric scaffolds with inflated assembly metrics if a true evolutionary relationship was overlooked. Annotation quality, however, has minimal effect on synteny if the assembled genome is highly contiguous. Conclusions Our results show that a minimum N50 of 1 Mb is required for robust downstream synteny analysis, which emphasizes the importance of gold standard genomes to the science community, and should be achieved given the current progress in sequencing technology.

D.6.8 Pseudogenes, new genes and small proteins

• Van Oss and Carvunis (2019) — A short, clear review of how genes arise from non-coding DNA, proto-genes, and the methodological pitfalls in identifying de novo genes. (review; CC BY 4.0)

  De novo gene birth. Van Oss & Carvunis (2019). PLOS Genetics. doi:10.1371/journal.pgen.1008160

• Wang et al. (2020) — A worked example that weighs the different routes to new genes in one lineage, showing how each is detected. (primary study; CC BY 4.0)

  De Novo Gene Birth, Horizontal Gene Transfer, and Gene Duplication as Sources of New Gene Families Associated with the Origin of Symbiosis inAmanita. Wang et al. (2020). Genome Biology and Evolution. doi:10.1093/gbe/evaa193

  Abstract

  By introducing novel capacities and functions, new genes and gene families may play a crucial role in ecological transitions. Mechanisms generating new gene families include de novo gene birth, horizontal gene transfer, and neofunctionalization following a duplication event. The ectomycorrhizal (ECM) symbiosis is a ubiquitous mutualism and the association has evolved repeatedly and independently many times among the fungi, but the evolutionary dynamics enabling its emergence remain elusive. We developed a phylogenetic workflow to first understand if gene families unique to ECM Amanita fungi and absent from closely related asymbiotic species are functionally relevant to the symbiosis, and then to systematically infer their origins. We identified 109 gene families unique to ECM Amanita species. Genes belonging to unique gene families are under strong purifying selection and are upregulated during symbiosis, compared with genes of conserved or orphan gene families. The origins of seven of the unique gene families are strongly supported as either de novo gene birth (two gene families), horizontal gene transfer (four), or gene duplication (one). An additional 34 families appear new because of their selective retention within symbiotic species. Among the 109 unique gene families, the most upregulated gene in symbiotic cultures encodes a 1-aminocyclopropane-1-carboxylate deaminase, an enzyme capable of downregulating the synthesis of the plant hormone ethylene, a common negative regulator of plant-microbial mutualisms.

D.7 Expression analysis

Further explanations for Chapter 7.

D.7.1 The first genome-wide methods: microarrays

• Lowe et al. (2017) — A short primer that compares microarrays and RNA-seq side by side, from sample preparation and normalisation to differential expression. Useful for seeing why RNA-seq replaced arrays and what the two technologies share. (primer; CC BY 4.0)

  Transcriptomics technologies. Lowe et al. (2017). PLOS Computational Biology. doi:10.1371/journal.pcbi.1005457

  Abstract

  Transcriptomics technologies are the techniques used to study an organism's transcriptome, the sum of all of its RNA transcripts. The information content of an organism is recorded in the DNA of its genome and expressed through transcription. Here, mRNA serves as a transient intermediary molecule in the information network, whilst noncoding RNAs perform additional diverse functions. A transcriptome captures a snapshot in time of the total transcripts present in a cell. The first attempts to study the whole transcriptome began in the early 1990s, and technological advances since the late 1990s have made transcriptomics a widespread discipline. Transcriptomics has been defined by repeated technological innovations that transform the field. There are two key contemporary techniques in the field: microarrays, which quantify a set of predetermined sequences, and RNA sequencing (RNA-Seq), which uses high-throughput sequencing to capture all sequences. Measuring the expression of an organism's genes in different tissues, conditions, or time points gives information on how genes are regulated and reveals details of an organism's biology. It can also help to infer the functions of previously unannotated genes. Transcriptomic analysis has enabled the study of how gene expression changes in different organisms and has been instrumental in the understanding of human disease. An analysis of gene expression in its entirety allows detection of broad coordinated trends which cannot be discerned by more targeted assays.

• Slonim and Yanai (2009) — An introductory tutorial on the microarray analysis workflow: quality control, normalisation, differential expression, clustering and functional interpretation. Older, but it describes the classical array pipeline that the section summarises. (tutorial; CC BY 4.0)

  Getting Started in Gene Expression Microarray Analysis. Slonim & Yanai (2009). PLoS Computational Biology. doi:10.1371/journal.pcbi.1000543

  Abstract

  Gene expression microarrays provide a snapshot of all the transcriptional activity in a biological sample. Unlike most traditional molecular biology tools, which generally allow the study of a single gene or a small set of genes, microarrays facilitate the discovery of totally novel and unexpected functional roles of genes. The power of these tools has been applied to a range of applications, including discovering novel disease subtypes, developing new diagnostic tools, and identifying underlying mechanisms of disease or drug response. However, this technology necessarily produces a large amount of data, challenging us to interpret it by exploiting modern computational and statistical tools. In this brief review, we aim to indicate the major issues involved in microarray analysis and provide a useful starting point for new microarray users. Figure 1 outlines the steps in a typical expression microarray experiment and maps them to the different sections of this review. Figure 1 Overview of steps in a typical gene expression microarray experiment.

D.7.2 RNA sequencing

• Law et al. (2018) — A complete worked analysis of a real RNA-seq data set in R, from count filtering and TMM normalisation to MDS plots, voom/limma testing and gene set tests. Shows in practice how the normalisation and modelling steps of the section fit together. (tutorial; CC BY 4.0)

  RNA-seq analysis is easy as 1-2-3 with limma, Glimma and edgeR. Law et al. (2018). F1000Research. doi:10.12688/f1000research.9005.3

  Abstract

  The ability to easily and efficiently analyse RNA-sequencing data is a key strength of the Bioconductor project. Starting with counts summarised at the gene-level, a typical analysis involves pre-processing, exploratory data analysis, differential expression testing and pathway analysis with the results obtained informing future experiments and validation studies. In this workflow article, we analyse RNA-sequencing data from the mouse mammary gland, demonstrating use of the popular edgeR package to import, organise, filter and normalise the data, followed by the limma package with its voom method, linear modelling and empirical Bayes moderation to assess differential expression and perform gene set testing. This pipeline is further enhanced by the Glimma package which enables interactive exploration of the results so that individual samples and genes can be examined by the user. The complete analysis offered by these three packages highlights the ease with which researchers can turn the raw counts from an RNA-sequencing experiment into biological insights using Bioconductor.

• Love et al. (2016) — A Bioconductor workflow from transcript quantification to DESeq2 testing, including size-factor normalisation, variance-stabilising transformations, exploratory PCA and batch handling. A useful complement to the limma/edgeR workflow above. (tutorial; CC BY 4.0)

  RNA-Seq workflow: gene-level exploratory analysis and differential expression. Love et al. (2016). F1000Research. doi:10.12688/f1000research.7035.2

  Abstract

  Here we walk through an end-to-end gene-level RNA-Seq differential expression workflow using Bioconductor packages. We will start from the FASTQ files, show how these were aligned to the reference genome, and prepare a count matrix which tallies the number of RNA-seq reads/fragments within each gene for each sample.We will perform exploratory data analysis (EDA) for quality assessment and to explore the relationship between samples, perform differential gene expression analysis, and visually explore the results.

D.7.3 Testing and statistics

• Korthauer et al. (2019) — Explains and benchmarks Bonferroni, Benjamini–Hochberg, Storey’s q-value and newer covariate-aware FDR methods on genomic examples. Helpful for understanding the multiple-testing part of the section and when the methods differ. (review; CC BY 4.0)

  A practical guide to methods controlling false discoveries in computational biology. Korthauer et al. (2019). Genome Biology. doi:10.1186/s13059-019-1716-1

  Abstract

  Background In high-throughput studies, hundreds to millions of hypotheses are typically tested. Statistical methods that control the false discovery rate (FDR) have emerged as popular and powerful tools for error rate control. While classic FDR methods use only p values as input, more modern FDR methods have been shown to increase power by incorporating complementary information as informative covariates to prioritize, weight, and group hypotheses. However, there is currently no consensus on how the modern methods compare to one another. We investigate the accuracy, applicability, and ease of use of two classic and six modern FDR-controlling methods by performing a systematic benchmark comparison using simulation studies as well as six case studies in computational biology. Results Methods that incorporate informative covariates are modestly more powerful than classic approaches, and do not underperform classic approaches, even when the covariate is completely uninformative. The majority of methods are successful at controlling the FDR, with the exception of two modern methods under certain settings. Furthermore, we find that the improvement of the modern FDR methods over the classic methods increases with the informativeness of the covariate, total number of hypothesis tests, and proportion of truly non-null hypotheses. Conclusions Modern FDR methods that use an informative covariate provide advantages over classic FDR-controlling procedures, with the relative gain dependent on the application and informativeness of available covariates. We present our findings as a practical guide and provide recommendations to aid researchers in their choice of methods to correct for false discoveries.

• Nguyen and Holmes (2019) — Short practical advice on PCA and related methods: preprocessing, choosing the number of dimensions, interpreting axes and checking stability. Supports the clustering and dimensionality-reduction subsection. (tutorial; CC BY 4.0)

  Ten quick tips for effective dimensionality reduction. Nguyen & Holmes (2019). PLOS Computational Biology. doi:10.1371/journal.pcbi.1006907

• Kobak and Berens (2019) — Explains how t-SNE works and how initialisation, perplexity and exaggeration affect the global structure of the embedding. Useful background for the warnings in the section about over-interpreting t-SNE and UMAP plots. (tutorial; CC BY 4.0)

  The art of using t-SNE for single-cell transcriptomics. Kobak & Berens (2019). Nature Communications. doi:10.1038/s41467-019-13056-x

  Abstract

  Single-cell transcriptomics yields ever growing data sets containing RNA expression levels for thousands of genes from up to millions of cells. Common data analysis pipelines include a dimensionality reduction step for visualising the data in two dimensions, most frequently performed using t-distributed stochastic neighbour embedding (t-SNE). It excels at revealing local structure in high-dimensional data, but naive applications often suffer from severe shortcomings, e.g. the global structure of the data is not represented accurately. Here we describe how to circumvent such pitfalls, and develop a protocol for creating more faithful t-SNE visualisations. It includes PCA initialisation, a high learning rate, and multi-scale similarity kernels; for very large data sets, we additionally use exaggeration and downsampling-based initialisation. We use published single-cell RNA-seq data sets to demonstrate that this protocol yields superior results compared to the naive application of t-SNE.

D.7.4 From gene lists to biology

• Wijesooriya et al. (2022) — Surveys how enrichment analysis is used in published studies and lists common errors, such as an inappropriate background gene list and missing correction for multiple testing. A practical guide to doing over-representation analysis correctly. (perspective; CC BY 4.0)

  Urgent need for consistent standards in functional enrichment analysis. Wijesooriya et al. (2022). PLOS Computational Biology. doi:10.1371/journal.pcbi.1009935

  Abstract

  Gene set enrichment tests (a.k.a. functional enrichment analysis) are among the most frequently used methods in computational biology. Despite this popularity, there are concerns that these methods are being applied incorrectly and the results of some peer-reviewed publications are unreliable. These problems include the use of inappropriate background gene lists, lack of false discovery rate correction and lack of methodological detail. To ascertain the frequency of these issues in the literature, we performed a screen of 186 open-access research articles describing functional enrichment results. We find that 95% of analyses using over-representation tests did not implement an appropriate background gene list or did not describe this in the methods. Failure to perform p-value correction for multiple tests was identified in 43% of analyses. Many studies lacked detail in the methods section about the tools and gene sets used. An extension of this survey showed that these problems are not associated with journal or article level bibliometrics. Using seven independent RNA-seq datasets, we show misuse of enrichment tools alters results substantially. In conclusion, most published functional enrichment studies suffered from one or more major flaws, highlighting the need for stronger standards for enrichment analysis.

• Dam et al. (2017) — Reviews how co-expression networks are built and used to predict gene function and disease genes, including guilt by association, WGCNA-type modules and data requirements. Covers the co-expression part of the section. (review; CC BY 4.0)

  Gene co-expression analysis for functional classification and gene–disease predictions. van Dam et al. (2017). Briefings in Bioinformatics. doi:10.1093/bib/bbw139

  Abstract

  Gene co-expression networks can be used to associate genes of unknown function with biological processes, to prioritize candidate disease genes or to discern transcriptional regulatory programmes. With recent advances in transcriptomics and next-generation sequencing, co-expression networks constructed from RNA sequencing data also enable the inference of functions and disease associations for non-coding genes and splice variants. Although gene co-expression networks typically do not provide information about causality, emerging methods for differential co-expression analysis are enabling the identification of regulatory genes underlying various phenotypes. Here, we introduce and guide researchers through a (differential) co-expression analysis. We provide an overview of methods and tools used to create and analyse co-expression networks constructed from gene expression data, and we explain how these can be used to identify genes with a regulatory role in disease. Furthermore, we discuss the integration of other data types with co-expression networks and offer future perspectives of co-expression analysis.

• Gibson et al. (2015) — Introduces cis- and trans-eQTL mapping and how eQTLs are used to interpret disease-associated variants from GWAS. Background for the regulatory-variation paragraph. (review; CC BY 4.0)

  Expression quantitative trait locus analysis for translational medicine. Gibson et al. (2015). Genome Medicine. doi:10.1186/s13073-015-0186-7

  Abstract

  Expression quantitative trait locus analysis has emerged as an important component of efforts to understand how genetic polymorphisms influence disease risk and is poised to make contributions to translational medicine. Here we review how expression quantitative trait locus analysis is aiding the identification of which gene(s) within regions of association are causal for a disease or phenotypic trait; the narrowing down of the cell types or regulators involved in the etiology of disease; the characterization of drivers and modifiers of cancer; and our understanding of how different environments and cellular contexts can modify gene expression. We also introduce the concept of transcriptional risk scores as a means of refining estimates of individual liability to disease based on targeted profiling of the transcripts that are regulated by polymorphisms jointly associated with disease and gene expression.

D.7.5 Single-cell genomics

• Haque et al. (2017) — An accessible overview of single-cell RNA-seq: cell isolation, plate- and droplet-based protocols, UMIs, and the main analysis steps. Good for the technologies and properties-of-the-data subsections; the analysis tools have since moved on. (review; CC BY 4.0)

  A practical guide to single-cell RNA-sequencing for biomedical research and clinical applications. Haque et al. (2017). Genome Medicine. doi:10.1186/s13073-017-0467-4

  Abstract

  RNA sequencing (RNA-seq) is a genomic approach for the detection and quantitative analysis of messenger RNA molecules in a biological sample and is useful for studying cellular responses. RNA-seq has fueled much discovery and innovation in medicine over recent years. For practical reasons, the technique is usually conducted on samples comprising thousands to millions of cells. However, this has hindered direct assessment of the fundamental unit of biology-the cell. Since the first single-cell RNA-sequencing (scRNA-seq) study was published in 2009, many more have been conducted, mostly by specialist laboratories with unique skills in wet-lab single-cell genomics, bioinformatics, and computation. However, with the increasing commercial availability of scRNA-seq platforms, and the rapid ongoing maturation of bioinformatics approaches, a point has been reached where any biomedical researcher or clinician can use scRNA-seq to make exciting discoveries. In this review, we present a practical guide to help researchers design their first scRNA-seq studies, including introductory information on experimental hardware, protocol choice, quality control, data analysis and biological interpretation.

• Williams et al. (2022) — Describes sequencing-based and imaging-based spatial transcriptomics methods, their resolution and gene coverage, and the computational analyses applied to them. Covers the spatial transcriptomics subsection. (review; CC BY 4.0)

  An introduction to spatial transcriptomics for biomedical research. Williams et al. (2022). Genome Medicine. doi:10.1186/s13073-022-01075-1

  Abstract

  Single-cell transcriptomics (scRNA-seq) has become essential for biomedical research over the past decade, particularly in developmental biology, cancer, immunology, and neuroscience. Most commercially available scRNA-seq protocols require cells to be recovered intact and viable from tissue. This has precluded many cell types from study and largely destroys the spatial context that could otherwise inform analyses of cell identity and function. An increasing number of commercially available platforms now facilitate spatially resolved, high-dimensional assessment of gene transcription, known as 'spatial transcriptomics'. Here, we introduce different classes of method, which either record the locations of hybridized mRNA molecules in tissue, image the positions of cells themselves prior to assessment, or employ spatial arrays of mRNA probes of pre-determined location. We review sizes of tissue area that can be assessed, their spatial resolution, and the number and types of genes that can be profiled. We discuss if tissue preservation influences choice of platform, and provide guidance on whether specific platforms may be better suited to discovery screens or hypothesis testing. Finally, we introduce bioinformatic methods for analysing spatial transcriptomic data, including pre-processing, integration with existing scRNA-seq data, and inference of cell-cell interactions. Spatial -omics methods are already improving our understanding of human tissues in research, diagnostic, and therapeutic settings. To build upon these recent advancements, we provide entry-level guidance for those seeking to employ spatial transcriptomics in their own biomedical research.

D.7.6 Single-cell sequencing and cancer evolution

• Vendramin et al. (2021) — Reviews cancer as an evolutionary process: clonal selection, intratumour heterogeneity and genetic as well as non-genetic sources of diversity. Background for the clonal-evolution subsection. (review; CC BY 4.0)

  Cancer evolution: Darwin and beyond. Vendramin et al. (2021). The EMBO Journal. doi:10.15252/embj.2021108389

  Abstract

  Clinical and laboratory studies over recent decades have established branched evolution as a feature of cancer. However, while grounded in somatic selection, several lines of evidence suggest a Darwinian model alone is insufficient to fully explain cancer evolution. First, the role of macroevolutionary events in tumour initiation and progression contradicts Darwin's central thesis of gradualism. Whole-genome doubling, chromosomal chromoplexy and chromothripsis represent examples of single catastrophic events which can drive tumour evolution. Second, neutral evolution can play a role in some tumours, indicating that selection is not always driving evolution. Third, increasing appreciation of the role of the ageing soma has led to recent generalised theories of age-dependent carcinogenesis. Here, we review these concepts and others, which collectively argue for a model of cancer evolution which extends beyond Darwin. We also highlight clinical opportunities which can be grasped through targeting cancer vulnerabilities arising from non-Darwinian patterns of evolution.

• Beerenwinkel et al. (2014) — Reviews population-genetic models of tumour growth and the computational inference of clonal composition and tumour phylogenies from sequencing data. Links the section to the phylogenetic methods of the book. (review; CC BY 4.0)

  Cancer Evolution: Mathematical Models and Computational Inference. Beerenwinkel et al. (2014). Systematic Biology. doi:10.1093/sysbio/syu081

  Abstract

  Cancer is a somatic evolutionary process characterized by the accumulation of mutations, which contribute to tumor growth, clinical progression, immune escape, and drug resistance development. Evolutionary theory can be used to analyze the dynamics of tumor cell populations and to make inference about the evolutionary history of a tumor from molecular data. We review recent approaches to modeling the evolution of cancer, including population dynamics models of tumor initiation and progression, phylogenetic methods to model the evolutionary relationship between tumor subclones, and probabilistic graphical models to describe dependencies among mutations. Evolutionary modeling helps to understand how tumors arise and will also play an increasingly important prognostic role in predicting disease progression and the outcome of medical interventions, such as targeted therapy.

• Navin (2015) — Reviews early single-cell DNA sequencing of tumours, including whole-genome amplification, copy-number profiling and the punctuated evolution observed in breast cancer. Directly relevant to the single-cell DNA sequencing subsection; link only because of the NC licence. (review; CC BY-NC (link only))

  The first five years of single-cell cancer genomics and beyond. Navin (2015). Genome Research. doi:10.1101/gr.191098.115

  Abstract

  Single-cell sequencing (SCS) is a powerful new tool for investigating evolution and diversity in cancer and understanding the role of rare cells in tumor progression. These methods have begun to unravel key questions in cancer biology that have been difficult to address with bulk tumor measurements. Over the past five years, there has been extraordinary progress in technological developments and research applications, but these efforts represent only the tip of the iceberg. In the coming years, SCS will greatly improve our understanding of invasion, metastasis, and therapy resistance during cancer progression. These tools will also have direct translational applications in the clinic, in areas such as early detection, noninvasive monitoring, and guiding targeted therapy. In this perspective, I discuss the progress that has been made and the myriad of unexplored applications that still lie ahead in cancer research and medicine.

D.8 Proteomics

Further explanations for Chapter 8.

D.8.1 Mass spectrometry methods

• Dupree et al. (2020) — Walks through the bottom-up workflow from sample preparation and digestion to LC-MS/MS, acquisition and quantification, and discusses its limitations. A readable overview of the section’s experimental side. (review; CC BY 4.0)

  A Critical Review of Bottom-Up Proteomics: The Good, the Bad, and the Future of This Field. Dupree et al. (2020). Proteomes. doi:10.3390/proteomes8030014

  Abstract

  Proteomics is the field of study that includes the analysis of proteins, from either a basic science prospective or a clinical one. Proteins can be investigated for their abundance, variety of proteoforms due to post-translational modifications (PTMs), and their stable or transient protein-protein interactions. This can be especially beneficial in the clinical setting when studying proteins involved in different diseases and conditions. Here, we aim to describe a bottom-up proteomics workflow from sample preparation to data analysis, including all of its benefits and pitfalls. We also describe potential improvements in this type of proteomics workflow for the future.

• Wang et al. (2024) — Tutorial on multiplex isotope labelling for quantification (precursor-based labels such as SILAC, mass-defect labels and reporter-ion labels such as TMT), with their principles, advantages and limitations such as ratio distortion. Useful background for the quantification subsection. (tutorial; CC BY 4.0)

  A Tutorial Review of Labeling Methods in Mass Spectrometry-Based Quantitative Proteomics. Wang et al. (2024). ACS Measurement Science Au. doi:10.1021/acsmeasuresciau.4c00007

  Abstract

  Recent advancements in mass spectrometry (MS) have revolutionized quantitative proteomics, with multiplex isotope labeling emerging as a key strategy for enhancing accuracy, precision, and throughput. This tutorial review offers a comprehensive overview of multiplex isotope labeling techniques, including precursor-based, mass defect-based, reporter ion-based, and hybrid labeling methods. It details their fundamental principles, advantages, and inherent limitations along with strategies to mitigate the limitation of ratio-distortion. This review will also cover the applications and latest progress in these labeling techniques across various domains, including cancer biomarker discovery, neuroproteomics, post-translational modification analysis, cross-linking MS, and single-cell proteomics. This Review aims to provide guidance for researchers on selecting appropriate methods for their specific goals while also highlighting the potential future directions in this rapidly evolving field.

• Jiang et al. (2024) — Long, comprehensive review covering instruments, fragmentation, acquisition modes, data analysis and quantification in bottom-up proteomics. Useful as a reference to dip into; licence allows linking only. (review; CC BY-NC-ND 4.0 (link only))

  Comprehensive Overview of Bottom-Up Proteomics Using Mass Spectrometry. Jiang et al. (2024). ACS Measurement Science Au. doi:10.1021/acsmeasuresciau.3c00068

  Abstract

  Proteomics is the large scale study of protein structure and function from biological systems through protein identification and quantification. "Shotgun proteomics" or "bottom-up proteomics" is the prevailing strategy, in which proteins are hydrolyzed into peptides that are analyzed by mass spectrometry. Proteomics studies can be applied to diverse studies ranging from simple protein identification to studies of proteoforms, protein-protein interactions, protein structural alterations, absolute and relative protein quantification, post-translational modifications, and protein stability. To enable this range of different experiments, there are diverse strategies for proteome analysis. The nuances of how proteomic workflows differ may be challenging to understand for new practitioners. Here, we provide a comprehensive overview of different proteomics methods. We cover from biochemistry basics and protein extraction to biological interpretation and orthogonal validation. We expect this Review will serve as a handbook for researchers who are new to the field of bottom-up proteomics.

D.8.2 Algorithms for peptide identification

• McHugh and Arthur (2008) — Short introductory review of peptide mass fingerprinting, database search, de novo sequencing and their scoring, written for computational readers. Older, but the basic algorithms it explains have not changed. (review; CC BY 4.0)

  Computational Methods for Protein Identification from Mass Spectrometry Data. McHugh & Arthur (2008). PLoS Computational Biology. doi:10.1371/journal.pcbi.0040012

  Abstract

  Protein identification using mass spectrometry is an indispensable computational tool in the life sciences. A dramatic increase in the use of proteomic strategies to understand the biology of living systems generates an ongoing need for more effective, efficient, and accurate computational methods for protein identification. A wide range of computational methods, each with various implementations, are available to complement different proteomic approaches. A solid knowledge of the range of algorithms available and, more critically, the accuracy and effectiveness of these techniques is essential to ensure as many of the proteins as possible, within any particular experiment, are correctly identified. Here, we undertake a systematic review of the currently available methods and algorithms for interpreting, managing, and analyzing biological data associated with protein identification. We summarize the advances in computational solutions as they have responded to corresponding advances in mass spectrometry hardware. The evolution of scoring algorithms and metrics for automated protein identification are also discussed with a focus on the relative performance of different techniques. We also consider the relative advantages and limitations of different techniques in particular biological contexts. Finally, we present our perspective on future developments in the area of computational protein identification by considering the most recent literature on new and promising approaches to the problem as well as identifying areas yet to be explored and the potential application of methods from other areas of computational biology.

• Angelis et al. (2025) — Introduction to deep-learning models that predict fragment spectra, retention times and other peptide properties, and how they are used in rescoring and library generation. Matches the subsection on predicting spectra and retention times. (tutorial; CC BY 4.0)

  Peptide Property Prediction for Mass Spectrometry Using AI: An Introduction to State of the Art Models. Angelis et al. (2025). PROTEOMICS. doi:10.1002/pmic.202400398

  Abstract

  This review explores state of the art machine learning and deep learning models for peptide property prediction in mass spectrometry-based proteomics, including, but not limited to, models for predicting digestibility, retention time, charge state distribution, collisional cross section, fragmentation ion intensities, and detectability. The combination of these models enables not only the in silico generation of spectral libraries but also finds many additional use cases in the design of targeted assays or data-driven rescoring. This review serves as both an introduction for newcomers and an update for experienced researchers aiming to develop accessible and reproducible models for peptide property predictions. Key limitations of the current models, including difficulties in handling diverse post-translational modifications and instrument variability, highlight the need for large-scale, harmonized datasets, and standardized evaluation metrics for benchmarking.

• Wen et al. (2020) — Overview of deep-learning applications in proteomics, including spectrum and retention-time prediction and de novo peptide sequencing. Gives a broader context for the machine-learning methods in this section. (review; CC BY 4.0)

  Deep Learning in Proteomics. Wen et al. (2020). PROTEOMICS. doi:10.1002/pmic.201900335

  Abstract

  Proteomics, the study of all the proteins in biological systems, is becoming a data-rich science. Protein sequences and structures are comprehensively catalogued in online databases. With recent advancements in tandem mass spectrometry (MS) technology, protein expression and post-translational modifications (PTMs) can be studied in a variety of biological systems at the global scale. Sophisticated computational algorithms are needed to translate the vast amount of data into novel biological insights. Deep learning automatically extracts data representations at high levels of abstraction from data, and it thrives in data-rich scientific research domains. Here, a comprehensive overview of deep learning applications in proteomics, including retention time prediction, MS/MS spectrum prediction, de novo peptide sequencing, PTM prediction, major histocompatibility complex-peptide binding prediction, and protein structure prediction, is provided. Limitations and the future directions of deep learning in proteomics are also discussed. This review will provide readers an overview of deep learning and how it can be used to analyze proteomics data.

D.8.3 Statistics and error control

• Noble and MacCoss (2012) — A chapter of the PLoS Computational Biology ‘Translational Bioinformatics’ collection that explains peptide identification, target–decoy FDR estimation, q-values, posterior error probabilities, protein inference and quantification. Written by researchers behind Percolator and Skyline, it is a clear statistical introduction to the whole section. (book chapter; CC BY 4.0)

  Computational and Statistical Analysis of Protein Mass Spectrometry Data. Noble & MacCoss (2012). PLoS Computational Biology. doi:10.1371/journal.pcbi.1002296

  Abstract

  High-throughput proteomics experiments involving tandem mass spectrometry produce large volumes of complex data that require sophisticated computational analyses. As such, the field offers many challenges for computational biologists. In this article, we briefly introduce some of the core computational and statistical problems in the field and then describe a variety of outstanding problems that readers of PLoS Computational Biology might be able to help solve.

• Wen et al. (2025) — Explains how entrapment experiments can test whether a pipeline’s reported FDR is valid, and shows that some widely used tools do not control the FDR as claimed. Useful for understanding the assumptions behind target–decoy estimates. (perspective; CC BY 4.0)

  Assessment of false discovery rate control in tandem mass spectrometry analysis using entrapment. Wen et al. (2025). Nature Methods. doi:10.1038/s41592-025-02719-x

  Abstract

  A critical challenge in mass spectrometry proteomics is accurately assessing error control, especially given that software tools employ distinct methods for reporting errors. Many tools are closed-source and poorly documented, leading to inconsistent validation strategies. Here we identify three prevalent methods for validating false discovery rate (FDR) control: one invalid, one providing only a lower bound, and one valid but under-powered. The result is that the proteomics community has limited insight into actual FDR control effectiveness, especially for data-independent acquisition (DIA) analyses. We propose a theoretical framework for entrapment experiments, allowing us to rigorously characterize different approaches. Moreover, we introduce a more powerful evaluation method and apply it alongside existing techniques to assess existing tools. We first validate our analysis in the better-understood data-dependent acquisition setup, and then, we analyze DIA data, where we find that no DIA search tool consistently controls the FDR, with particularly poor performance on single-cell datasets.

• Hutchings et al. (2023) — Step-by-step Bioconductor workflow for TMT and label-free data, covering import, pre-processing, quality control, aggregation to proteins and differential abundance testing. A practical complement to the statistics of quantification. (tutorial; CC BY 4.0)

  A Bioconductor workflow for processing, evaluating, and interpreting expression proteomics data. Hutchings et al. (2023). F1000Research. doi:10.12688/f1000research.139116.1

  Abstract

  Background: Expression proteomics involves the global evaluation of protein abundances within a system. In turn, differential expression analysis can be used to investigate changes in protein abundance upon perturbation to such a system. Methods: Here, we provide a workflow for the processing, analysis and interpretation of quantitative mass spectrometry-based expression proteomics data. This workflow utilizes open-source R software packages from the Bioconductor project and guides users end-to-end and step-by-step through every stage of the analyses. As a use-case we generated expression proteomics data from HEK293 cells with and without a treatment. Of note, the experiment included cellular proteins labelled using tandem mass tag (TMT) technology and secreted proteins quantified using label-free quantitation (LFQ). Results: The workflow explains the software infrastructure before focusing on data import, pre-processing and quality control. This is done individually for TMT and LFQ datasets. The application of statistical differential expression analysis is demonstrated, followed by interpretation via gene ontology enrichment analysis. Conclusions: A comprehensive workflow for the processing, analysis and interpretation of expression proteomics is presented. The workflow is a valuable resource for the proteomics community and specifically beginners who are at least familiar with R who wish to understand and make data-driven decisions with regards to their analyses.

D.8.4 Proteomics beyond protein abundance

• Richards et al. (2021) — Reviews affinity purification–MS, proximity labelling, co-fractionation and cross-linking MS for mapping interaction networks, including scoring and quality control. Covers the interaction proteomics subsection. (review; CC BY 4.0)

  Mass spectrometry-based protein–protein interaction networks for the study of human diseases. Richards et al. (2021). Molecular Systems Biology. doi:10.15252/msb.20188792

  Abstract

  A better understanding of the molecular mechanisms underlying disease is key for expediting the development of novel therapeutic interventions. Disease mechanisms are often mediated by interactions between proteins. Insights into the physical rewiring of protein-protein interactions in response to mutations, pathological conditions, or pathogen infection can advance our understanding of disease etiology, progression, and pathogenesis and can lead to the identification of potential druggable targets. Advances in quantitative mass spectrometry (MS)-based approaches have allowed unbiased mapping of these disease-mediated changes in protein-protein interactions on a global scale. Here, we review MS techniques that have been instrumental for the identification of protein-protein interactions at a system-level, and we discuss the challenges associated with these methodologies as well as novel MS advancements that aim to address these challenges. An overview of examples from diverse disease contexts illustrates the potential of MS-based protein-protein interaction mapping approaches for revealing disease mechanisms, pinpointing new therapeutic targets, and eventually moving toward personalized applications.

• Savage and Zhang (2020) — Guide to bioinformatics resources for phosphoproteomics: phosphosite and kinase knowledge bases, prediction of sites and their kinases, and kinase activity inference. Useful for the post-translational modifications subsection. (review; CC BY 4.0)

  Using phosphoproteomics data to understand cellular signaling: a comprehensive guide to bioinformatics resources. Savage & Zhang (2020). Clinical Proteomics. doi:10.1186/s12014-020-09290-x

  Abstract

  Mass spectrometry-based phosphoproteomics is becoming an essential methodology for the study of global cellular signaling. Numerous bioinformatics resources are available to facilitate the translation of phosphopeptide identification and quantification results into novel biological and clinical insights, a critical step in phosphoproteomics data analysis. These resources include knowledge bases of kinases and phosphatases, phosphorylation sites, kinase inhibitors, and sequence variants affecting kinase function, and bioinformatics tools that can predict phosphorylation sites in addition to the kinase that phosphorylates them, infer kinase activity, and predict the effect of mutations on kinase signaling. However, these resources exist in silos and it is challenging to select among multiple resources with similar functions. Therefore, we put together a comprehensive collection of resources related to phosphoproteomics data interpretation, compared the use of tools with similar functions, and assessed the usability from the standpoint of typical biologists or clinicians. Overall, tools could be improved by standardization of enzyme names, flexibility of data input and output format, consistent maintenance, and detailed manuals.

• Masson et al. (2019) — Community recommendations that also explain the principles of HDX-MS, how the data are analysed and how differences between states are judged statistically. A good entry point to structural mass spectrometry. (perspective; CC BY 4.0)

  Recommendations for performing, interpreting and reporting hydrogen deuterium exchange mass spectrometry (HDX-MS) experiments. Masson et al. (2019). Nature Methods. doi:10.1038/s41592-019-0459-y

  Abstract

  Hydrogen deuterium exchange mass spectrometry (HDX-MS) is a powerful biophysical technique being increasingly applied to a wide variety of problems. As the HDX-MS community continues to grow, adoption of best practices in data collection, analysis, presentation and interpretation will greatly enhance the accessibility of this technique to nonspecialists. Here we provide recommendations arising from community discussions emerging out of the first International Conference on Hydrogen-Exchange Mass Spectrometry (IC-HDX; 2017). It is meant to represent both a consensus viewpoint and an opportunity to stimulate further additions and refinements as the field advances.

D.8.5 Proteomics without mass spectrometry

• Edfors et al. (2018) — Applies the ‘pillars’ of antibody validation (genetic, orthogonal, independent antibodies, tagged proteins, capture MS) to thousands of Human Protein Atlas antibodies. Explains why antibody-based proteomics needs systematic validation; it is a primary study rather than a review. (review; CC BY 4.0)

  Enhanced validation of antibodies for research applications. Edfors et al. (2018). Nature Communications. doi:10.1038/s41467-018-06642-y

  Abstract

  There is a need for standardized validation methods for antibody specificity and selectivity. Recently, five alternative validation pillars were proposed to explore the specificity of research antibodies using methods with no need for prior knowledge about the protein target. Here, we show that these principles can be used in a streamlined manner for enhanced validation of research antibodies in Western blot applications. More than 6,000 antibodies were validated with at least one of these strategies involving orthogonal methods, genetic knockdown, recombinant expression, independent antibodies, and capture mass spectrometry analysis. The results show a path forward for efforts to validate antibodies in an application-specific manner suitable for both providers and users.

• Eldjarn et al. (2023) — Compares the two main affinity platforms for plasma proteomics, SomaScan (aptamers) and Olink (proximity extension assay), in the same individuals using precision, correlation and genetic associations. A primary study, but the most direct explanation of how the two platforms differ. (review; CC BY 4.0)

  Large-scale plasma proteomics comparisons through genetics and disease associations. Eldjarn et al. (2023). Nature. doi:10.1038/s41586-023-06563-x

  Abstract

  High-throughput proteomics platforms measuring thousands of proteins in plasma combined with genomic and phenotypic information have the power to bridge the gap between the genome and diseases. Here we performed association studies of Olink Explore 3072 data generated by the UK Biobank Pharma Proteomics Project 1 on plasma samples from more than 50,000 UK Biobank participants with phenotypic and genotypic data, stratifying on British or Irish, African and South Asian ancestries. We compared the results with those of a SomaScan v4 study on plasma from 36,000 Icelandic people 2 , for 1,514 of whom Olink data were also available. We found modest correlation between the two platforms. Although cis protein quantitative trait loci were detected for a similar absolute number of assays on the two platforms (2,101 on Olink versus 2,120 on SomaScan), the proportion of assays with such supporting evidence for assay performance was higher on the Olink platform (72% versus 43%). A considerable number of proteins had genomic associations that differed between the platforms. We provide examples where differences between platforms may influence conclusions drawn from the integration of protein levels with the study of diseases. We demonstrate how leveraging the diverse ancestries of participants in the UK Biobank helps to detect novel associations and refine genomic location. Our results show the value of the information provided by the two most commonly used high-throughput proteomics platforms and demonstrate the differences between them that at times provides useful complementarity.

• Timp and Timp (2020) — Surveys alternatives to mass spectrometry for identifying and sequencing single protein molecules, such as nanopores and fluorescence-based methods, and their prospects. Background for the subsection on protein sequencing. (review; CC BY 4.0)

  Beyond mass spectrometry, the next step in proteomics. Timp & Timp (2020). Science Advances. doi:10.1126/sciadv.aax8978

  Abstract

  Proteins can be the root cause of a disease, and they can be used to cure it. The need to identify these critical actors was recognized early (1951) by Sanger; the first biopolymer sequenced was a peptide, insulin. With the advent of scalable, single-molecule DNA sequencing, genomics and transcriptomics have since propelled medicine through improved sensitivity and lower costs, but proteomics has lagged behind. Currently, proteomics relies mainly on mass spectrometry (MS), but instead of truly sequencing, it classifies a protein and typically requires about a billion copies of a protein to do it. Here, we offer a survey that illuminates a few alternatives with the brightest prospects for identifying whole proteins and displacing MS for sequencing them. These alternatives all boast sensitivity superior to MS and promise to be scalable and seem to be adaptable to bioinformatics tools for calling the sequence of amino acids that constitute a protein.

D.8.6 Repositories and standards

• Perez-Riverol et al. (2015) — Survey of proteomics repositories and knowledge bases, including PRIDE, PeptideAtlas and ProteomeXchange, and of how their data can be reused. Somewhat dated in details but a good map of the landscape. (review; CC BY 4.0)

  Making proteomics data accessible and reusable: Current state of proteomics databases and repositories. Perez-Riverol et al. (2015). PROTEOMICS. doi:10.1002/pmic.201400302

  Abstract

  Compared to other data-intensive disciplines such as genomics, public deposition and storage of MS-based proteomics, data are still less developed due to, among other reasons, the inherent complexity of the data and the variety of data types and experimental workflows. In order to address this need, several public repositories for MS proteomics experiments have been developed, each with different purposes in mind. The most established resources are the Global Proteome Machine Database (GPMDB), PeptideAtlas, and the PRIDE database. Additionally, there are other useful (in many cases recently developed) resources such as ProteomicsDB, Mass Spectrometry Interactive Virtual Environment (MassIVE), Chorus, MaxQB, PeptideAtlas SRM Experiment Library (PASSEL), Model Organism Protein Expression Database (MOPED), and the Human Proteinpedia. In addition, the ProteomeXchange consortium has been recently developed to enable better integration of public repositories and the coordinated sharing of proteomics information, maximizing its benefit to the scientific community. Here, we will review each of the major proteomics resources independently and some tools that enable the integration, mining and reuse of the data. We will also discuss some of the major challenges and current pitfalls in the integration and sharing of the data.

• Deutsch et al. (2023) — Describes the HUPO-PSI standards (mzML, mzIdentML, mzTab, Universal Spectrum Identifier, SDRF-Proteomics and others) and why they exist. Directly covers the standards part of the section. (review; CC BY 4.0)

  Proteomics Standards Initiative at Twenty Years: Current Activities and Future Work. Deutsch et al. (2023). Journal of Proteome Research. doi:10.1021/acs.jproteome.2c00637

  Abstract

  The Human Proteome Organization (HUPO) Proteomics Standards Initiative (PSI) has been successfully developing guidelines, data formats, and controlled vocabularies (CVs) for the proteomics community and other fields supported by mass spectrometry since its inception 20 years ago. Here we describe the general operation of the PSI, including its leadership, working groups, yearly workshops, and the document process by which proposals are thoroughly and publicly reviewed in order to be ratified as PSI standards. We briefly describe the current state of the many existing PSI standards, some of which remain the same as when originally developed, some of which have undergone subsequent revisions, and some of which have become obsolete. Then the set of proposals currently being developed are described, with an open call to the community for participation in the forging of the next generation of standards. Finally, we describe some synergies and collaborations with other organizations and look to the future in how the PSI will continue to promote the open sharing of data and thus accelerate the progress of the field of proteomics.

• Perez-Riverol et al. (2019) — Analyses how public proteomics and other omics data sets are downloaded and reused. Illustrates why deposition in repositories such as PRIDE matters. (perspective; CC BY 4.0)

  Quantifying the impact of public omics data. Perez-Riverol et al. (2019). Nature Communications. doi:10.1038/s41467-019-11461-w

  Abstract

  The amount of omics data in the public domain is increasing every year. Modern science has become a data-intensive discipline. Innovative solutions for data management, data sharing, and for discovering novel datasets are therefore increasingly required. In 2016, we released the first version of the Omics Discovery Index (OmicsDI) as a light-weight system to aggregate datasets across multiple public omics data resources. OmicsDI aggregates genomics, transcriptomics, proteomics, metabolomics and multiomics datasets, as well as computational models of biological processes. Here, we propose a set of novel metrics to quantify the attention and impact of biomedical datasets. A complete framework (now integrated into OmicsDI) has been implemented in order to provide and evaluate those metrics. Finally, we propose a set of recommendations for authors, journals and data resources to promote an optimal quantification of the impact of datasets.

D.9 Protein structure

Further explanations for Chapter 9.

D.9.1 CASP: a blind test of structure prediction

• Elofsson (2023) — A short review of what CASP15 showed after AlphaFold2: where monomer prediction stands, the progress for multimers and the categories that remain difficult. Useful as a compact account of how CASP results are read. (review; CC BY 4.0)

  Progress at protein structure prediction, as seen in CASP15. Elofsson (2023). Current Opinion in Structural Biology. doi:10.1016/j.sbi.2023.102594

  Abstract

  In Dec 2020, the results of AlphaFold version 2 were presented at CASP14, sparking a revolution in the field of protein structure predictions. For the first time, a purely computational method could challenge experimental accuracy for structure prediction of single protein domains. The code of AlphaFold v2 was released in the summer of 2021, and since then, it has been shown that it can be used to accurately predict the structure of most ordered proteins and many protein-protein interactions. It has also sparked an explosion of development in the field, improving AI-based methods to predict protein complexes, disordered regions, and protein design. Here I will review some of the inventions sparked by the release of AlphaFold.

• Abriata and Dal Peraro (2025) — Summarises the CASP16 results from the point of view of a user: which kinds of targets (monomers, complexes, ligands, nucleic acids) can be modelled reliably and which cannot. Helps to translate CASP metrics into practical expectations. (review; CC BY 4.0)

  Practical Outcomes From CASP16 for Users in Need of Biomolecular Structure Prediction. Abriata & Dal Peraro (2025). Proteins: Structure, Function, and Bioinformatics. doi:10.1002/prot.70078

  Abstract

  The 16th Critical Assessment of Structure Prediction benchmarked advancements in biomolecular modeling, particularly in the context of AlphaFold 2 and 3 systems. Protein monomer and domain prediction is largely solved, with barely any space for further improvements at the backbone level except for very specific details, irregular secondary structures, and mutational effects that remain challenging to predict. For protein assemblies, AF-based methods, especially when expertly guided or enhanced by servers like those from the Yang, Zheng/Zhang, and Cheng lab, show progress, though complex topologies and in particular antibody-antigen interactions are still difficult. Notably, a priori knowledge of stoichiometry significantly aids assembly prediction. Protein-ligand co-folding with AF3 demonstrated strong potential for pose prediction, outperforming many participants and some dedicated docking tools in baseline tests, but several caveats hold as discussed. Ligand affinity prediction is totally unreliable. Nucleic acid structure prediction lags considerably, heavily relying on 3D templates and expert human intervention, even AF3 showing substantial limitations. Overall, on all fronts, AF3's modeling capabilities are at or close to the state of the art; additionally, it shows slight improvements over AF2 and more detailed confidence metrics than it. We guide users on tool selection, realistic accuracy expectations, and persistent challenges, emphasizing the critical role of confidence metrics in interpreting AI-generated models.

D.9.2 Background: amino acids, energetics and enzymes

• Jakubowski & Flatt (2024) — Open textbook chapter on the chemistry of the amino acids, side-chain properties and the peptide bond; the neighbouring chapters of the same book cover protein structure and enzyme activity. Good for revising the biochemistry this section assumes. (book chapter; CC BY-NC-SA 4.0)

  Amino Acids, Peptides, and Proteins (Chapter 3 of Fundamentals of Biochemistry). Jakubowski & Flatt (2024). LibreTexts Biology (open textbook). https://bio.libretexts.org/Bookshelves/Biochemistry/Fundamentals_of_Biochemistry_(Jakubowski_and_Flatt)/01%3A_Unit_I-_Structure_and_Catalysis/03%3A_Amino_Acids_Peptides_and_Proteins

• Jarmoskaite et al. (2020) — Explains what a dissociation constant means and the controls needed to measure it correctly, with many examples of common errors. Gives a concrete feel for the thermodynamics of binding discussed in the section. (tutorial; CC BY 4.0)

  How to measure and evaluate binding affinities. Jarmoskaite et al. (2020). eLife. doi:10.7554/elife.57264

  Abstract

  Quantitative measurements of biomolecule associations are central to biological understanding and are needed to build and test predictive and mechanistic models. Given the advances in high-throughput technologies and the projected increase in the availability of binding data, we found it especially timely to evaluate the current standards for performing and reporting binding measurements. A review of 100 studies revealed that in most cases essential controls for establishing the appropriate incubation time and concentration regime were not documented, making it impossible to determine measurement reliability. Moreover, several reported affinities could be concluded to be incorrect, thereby impacting biological interpretations. Given these challenges, we provide a framework for a broad range of researchers to evaluate, teach about, perform, and clearly document high-quality equilibrium binding measurements. We apply this framework and explain underlying fundamental concepts through experimental examples with the RNA-binding protein Puf4.

• Cornish-Bowden (2015) — A short historical and conceptual account of the Michaelis–Menten equation, what kcat and Km mean and how they are often misinterpreted. Useful background for the enzyme part of the section. (perspective; CC BY 4.0)

  One hundred years of Michaelis–Menten kinetics. Cornish-Bowden (2015). Perspectives in Science. doi:10.1016/j.pisc.2014.12.002

  Abstract

  The year 2013 marked the centenary of the paper of Leonor Michaelis and Maud Menten (Michaelis and Menten, 1913), and the 110th anniversary of the doctoral thesis of Victor Henri (Henri, 1903). These publications have had an enormous influence on the progress of biochemistry, and are more often cited in the 21st century than they were in the 20th. Henri laid the groundwork for the understanding of enzyme mechanisms, but his experimental design was open to criticism. He reached essentially correct conclusions about the action of invertase, but he took no steps to control the hydrogen-ion concentration, and he took no account of the spontaneous mutarotation of the glucose produced in the reaction. Michaelis and Menten corrected these shortcomings, and in addition they introduced the initial-rate method of analysis, which has proved much simpler to apply than the methods based on time courses that it replaced. In this way they defined the methodology for steady-state experiments that has remained standard for 100 years.

D.9.3 Protein structure and how it is determined

• Vénien-Bryan et al. (2017) — Compares the principles, strengths and limitations of X-ray crystallography and single-particle cryo-EM, written for readers who are not specialists in either. Helps to understand why resolution and sample requirements differ between the methods. (review; CC BY 2.0 UK)

  Cryo-electron microscopy and X-ray crystallography: complementary approaches to structural biology and drug discovery. Vénien-Bryan et al. (2017). Acta Crystallographica Section F Structural Biology Communications. doi:10.1107/s2053230x17003740

  Abstract

  The invention of the electron microscope has greatly enhanced the view scientists have of small structural details. Since its implementation, this technology has undergone considerable evolution and the resolution that can be obtained for biological objects has been extended. In addition, the latest generation of cryo-electron microscopes equipped with direct electron detectors and software for the automated collection of images, in combination with the use of advanced image-analysis methods, has dramatically improved the performance of this technique in terms of resolution. While calculating a sub-10 Å resolution structure was an accomplishment less than a decade ago, it is now common to generate structures at sub-5 Å resolution and even better. It is becoming possible to relatively quickly obtain high-resolution structures of biological molecules, in particular large ones (>500 kDa) which, in some cases, have resisted more conventional methods such as X-ray crystallography or nuclear magnetic resonance (NMR). Such newly resolved structures may, for the first time, shed light on the precise mechanisms that are essential for cellular physiological processes. The ability to attain atomic resolution may support the development of new drugs that target these proteins, allowing medicinal chemists to understand the intimacy of the relationship between their molecules and targets. In addition, recent developments in cryo-electron microscopy combined with image analysis can provide unique information on the conformational variability of macromolecular complexes. Conformational flexibility of macromolecular complexes can be investigated using cryo-electron microscopy and multiconformation reconstruction methods. However, the biochemical quality of the sample remains the major bottleneck to routine cryo-electron microscopy-based determination of structures at very high resolution.

• Laurents (2022) — Reviews what NMR spectroscopy measures (structure, dynamics, disorder) and how it complements AlphaFold2 models. Useful for the NMR subsection and for seeing what a static prediction does not capture. (review; CC BY 4.0)

  AlphaFold 2 and NMR Spectroscopy: Partners to Understand Protein Structure, Dynamics and Function. Laurents (2022). Frontiers in Molecular Biosciences. doi:10.3389/fmolb.2022.906437

  Abstract

  The artificial intelligence program AlphaFold 2 is revolutionizing the field of protein structure determination as it accurately predicts the 3D structure of two thirds of the human proteome. Its predictions can be used directly as structural models or indirectly as aids for experimental structure determination using X-ray crystallography, CryoEM or NMR spectroscopy. Nevertheless, AlphaFold 2 can neither afford insight into how proteins fold, nor can it determine protein stability or dynamics. Rare folds or minor alternative conformations are also not predicted by AlphaFold 2 and the program does not forecast the impact of post translational modifications, mutations or ligand binding. The remaining third of human proteome which is poorly predicted largely corresponds to intrinsically disordered regions of proteins. Key to regulation and signaling networks, these disordered regions often form biomolecular condensates or amyloids. Fortunately, the limitations of AlphaFold 2 are largely complemented by NMR spectroscopy. This experimental approach provides information on protein folding and dynamics as well as biomolecular condensates and amyloids and their modulation by experimental conditions, small molecules, post translational modifications, mutations, flanking sequence, interactions with other proteins, RNA and virus. Together, NMR spectroscopy and AlphaFold 2 can collaborate to advance our comprehension of proteins.

• Terwilliger et al. (2023) — Compares AlphaFold models directly with experimental electron-density maps and shows where even high-confidence predictions differ from the data. Directly relevant to the subsection on experimental structures versus predicted models. (perspective; CC BY 4.0)

  AlphaFold predictions are valuable hypotheses and accelerate but do not replace experimental structure determination. Terwilliger et al. (2023). Nature Methods. doi:10.1038/s41592-023-02087-4

  Abstract

  Artificial intelligence-based protein structure prediction methods such as AlphaFold have revolutionized structural biology. The accuracies of these predictions vary, however, and they do not take into account ligands, covalent modifications or other environmental factors. Here, we evaluate how well AlphaFold predictions can be expected to describe the structure of a protein by comparing predictions directly with experimental crystallographic maps. In many cases, AlphaFold predictions matched experimental maps remarkably closely. In other cases, even very high-confidence predictions differed from experimental maps on a global scale through distortion and domain orientation, and on a local scale in backbone and side-chain conformation. We suggest considering AlphaFold predictions as exceptionally useful hypotheses. We further suggest that it is important to consider the confidence in prediction when interpreting AlphaFold predictions and to carry out experimental structure determination to verify structural details, particularly those that involve interactions not included in the prediction.

D.9.4 Protein folding

• Gianni et al. (2021) — Introduces the energy-landscape view of folding, the principle of minimal frustration and how residual frustration relates to function and disorder. A readable route from folding funnels to current ideas. (review; CC BY 4.0)

  Fuzziness and Frustration in the Energy Landscape of Protein Folding, Function, and Assembly. Gianni et al. (2021). Accounts of Chemical Research. doi:10.1021/acs.accounts.0c00813

  Abstract

  Are all protein interactions fully optimized? Do suboptimal interactions compromise specificity? What is the functional impact of frustration? Why does evolution not optimize some contacts? Proteins and their complexes are best described as ensembles of states populating an energy landscape. These ensembles vary in breadth from narrow ensembles clustered around a single average X-ray structure to broader ensembles encompassing a few different functional "taxonomic" states on to near continua of rapidly interconverting conformations, which are called "fuzzy" or even "intrinsically disordered". Here we aim to provide a comprehensive framework for confronting the structural and dynamical continuum of protein assemblies by combining the concepts of energetic frustration and interaction fuzziness. The diversity of the protein structural ensemble arises from the frustrated conflicts between the interactions that create the energy landscape. When frustration is minimal after folding, it results in a narrow ensemble, but residual frustrated interactions result in fuzzy ensembles, and this fuzziness allows a versatile repertoire of biological interactions. Here we discuss how fuzziness and frustration play off each other as proteins fold and assemble, viewing their significance from energetic, functional, and evolutionary perspectives.We demonstrate, in particular, that the common physical origin of both concepts is related to the ruggedness of the energy landscapes, intramolecular in the case of frustration and intermolecular in the case of fuzziness. Within this framework, we show that alternative sets of suboptimal contacts may encode specificity without achieving a single structural optimum. Thus, we demonstrate that structured complexes may not be optimized, and energetic frustration is realized via different sets of contacts leading to multiplicity of specific complexes. Furthermore, we propose that these suboptimal, frustrated, or fuzzy interactions are under evolutionary selection and expand the biological repertoire by providing a multiplicity of biological activities. In accord, we show that non-native interactions in folding or interaction landscapes can cooperate to generate diverse functional states, which are essential to facilitate adaptation to different cellular conditions. Thus, we propose that not fully optimized structures may actually be beneficial for biological activities of proteins via an alternative set of suboptimal interactions. The importance of such variability has not been recognized across different areas of biology.This account provides a modern view on folding, function, and assembly across the protein universe. The physical framework presented here is applicable to the structure and dynamics continuum of proteins and opens up new perspectives for drug design involving not fully structured, highly dynamic protein assemblies.

• Bowman (2024) — A short review arguing that predicting a single structure does not solve the folding problem, and that folding pathways and conformational ensembles remain open questions. Fits the subsection on folding and prediction. (review; CC BY 4.0)

  AlphaFold and Protein Folding: Not Dead Yet! The Frontier Is Conformational Ensembles. Bowman (2024). Annual Review of Biomedical Data Science. doi:10.1146/annurev-biodatasci-102423-011435

  Abstract

  Like the black knight in the classic Monty Python movie, grand scientific challenges such as protein folding are hard to finish off. Notably, AlphaFold is revolutionizing structural biology by bringing highly accurate structure prediction to the masses and opening up innumerable new avenues of research. Despite this enormous success, calling structure prediction, much less protein folding and related problems, "solved" is dangerous, as doing so could stymie further progress. Imagine what the world would be like if we had declared flight solved after the first commercial airlines opened and stopped investing in further research and development. Likewise, there are still important limitations to structure prediction that we would benefit from addressing. Moreover, we are limited in our understanding of the enormous diversity of different structures a single protein can adopt (called a conformational ensemble) and the dynamics by which a protein explores this space. What is clear is that conformational ensembles are critical to protein function, and understanding this aspect of protein dynamics will advance our ability to design new proteins and drugs.

D.9.5 Structure comparison and evolution

• Sahakyan et al. (2025) — A brief review of methods for detecting distant relationships between proteins, from sequence profiles to structure comparison with large predicted-structure databases. Shows why structure comparison reaches further back in evolution than sequence comparison. (review; CC BY 4.0)

  Exploring the protein universe with distant similarity detection methods. Sahakyan et al. (2025). Protein Science. doi:10.1002/pro.70397

  Abstract

  During the last few years, the body of data on proteins is expanding almost exponentially with the development of advanced methods for gene sequencing, protein structure determination, particularly by cryoelectron microscopy, and structure prediction using artificial intelligence-based approaches. These developments create the potential for a comprehensive exploration of the protein universe, the entirety of the proteins existing in the biosphere. Elucidation of the relationships among proteins including the most distant ones, where only the core fold is shared, is crucial for understanding protein functions, folding mechanisms, and evolution, as well as the evolution of cellular life forms and viruses. In this brief review, we discuss methods that shaped the field of protein bioinformatics, first, through comparative sequence analysis, and the recent developments in protein structure prediction that transformed the state of the art in the comparative analysis of distantly related proteins. The combination of the rapidly growing databases of genome and metagenome sequences with sensitive methods for sequence comparison and the new generation of structure analysis tools can make charting the protein universe at the structural level a realistic goal.

• Kulkarni et al. (2025) — A perspective on how views of the protein structure universe have changed, including fold switching, disorder and amyloid. Useful as background for what structure comparison can and cannot reveal about evolution. (perspective; CC BY 4.0)

  Evolving concepts of the protein universe. Kulkarni et al. (2025). iScience. doi:10.1016/j.isci.2025.112012

  Abstract

  The protein universe is the collection of all proteins on earth from all organisms both extant and extinct. Classical studies on protein folding suggested that proteins exist as a unique three-dimensional conformation that is dictated by the genetic code and is critical for function. In this perspective, we discuss ideas and developments that emerged over the past three decades regarding the protein structure-function paradigm. It is now clear that ordered (active/functional) and disordered/denatured (and hence inactive/non-functional) represent a continuum of states rather than binary states. Some proteins can switch folds without sequence change. Others exist as conformational ensembles lacking defined structure yet play critical roles in many biological processes, including forming membrane-less organelles driven by liquid-liquid phase separation. Numerous diverse proteins harbor segments with the potential to form amyloid fibrils, many of which are functional, and some possess prion-like properties enabling conformation-based transfer of heritable information. Taken together, these developments reveal the remarkable complexity of the protein universe.

D.9.6 Protein structure databases

• Burley et al. (2022) — Describes what the PDB contains, how entries are deposited and validated, and how the archive is used. A good overview of the central structure database. (review; CC BY 4.0)

  Protein Data Bank: A Comprehensive Review of 3D Structure Holdings and Worldwide Utilization by Researchers, Educators, and Students. Burley et al. (2022). Biomolecules. doi:10.3390/biom12101425

  Abstract

  The Research Collaboratory for Structural Bioinformatics Protein Data Bank (RCSB PDB), funded by the United States National Science Foundation, National Institutes of Health, and Department of Energy, supports structural biologists and Protein Data Bank (PDB) data users around the world. The RCSB PDB, a founding member of the Worldwide Protein Data Bank (wwPDB) partnership, serves as the US data center for the global PDB archive housing experimentally-determined three-dimensional (3D) structure data for biological macromolecules. As the wwPDB-designated Archive Keeper, RCSB PDB is also responsible for the security of PDB data and weekly update of the archive. RCSB PDB serves tens of thousands of data depositors (using macromolecular crystallography, nuclear magnetic resonance spectroscopy, electron microscopy, and micro-electron diffraction) annually working on all permanently inhabited continents. RCSB PDB makes PDB data available from its research-focused web portal at no charge and without usage restrictions to many millions of PDB data consumers around the globe. It also provides educators, students, and the general public with an introduction to the PDB and related training materials through its outreach and education-focused web portal. This review article describes growth of the PDB, examines evolution of experimental methods for structure determination viewed through the lens of the PDB archive, and provides a detailed accounting of PDB archival holdings and their utilization by researchers, educators, and students worldwide.

• Fleming et al. (2025) — Describes the contents and use of the AlphaFold Protein Structure Database and the 3D-Beacons network that links predicted and experimental models. Useful for understanding how predicted structures are distributed and accessed. (review; CC BY 4.0)

  AlphaFold Protein Structure Database and 3D-Beacons: New Data and Capabilities. Fleming et al. (2025). Journal of Molecular Biology. doi:10.1016/j.jmb.2025.168967

  Abstract

  The AlphaFold Protein Structure Database (https://alphafold.ebi.ac.uk/) has made significant strides in enhancing its utility and accessibility for the life science research community. The recent integration of AlphaMissense predictions enables access to the pathogenicity of human protein missense variants, with an innovative and interactive heatmap and 3D visualisation that display variant data at the residue level. Users can now toggle between structure model quality (pLDDT) and average pathogenicity scores, providing insights into the implications of specific residue changes. The Foldseek integration offers a rapid and accurate method for protein structure searches and comparisons. Bulk data download options further facilitate comprehensive data analysis and integration with other computational tools. The 3D-Beacons framework (https://www.ebi.ac.uk/pdbe/pdbe-kb/3dbeacons/) has also been enhanced with detailed annotation endpoints (such as AlphaMissense data) and integrates LevyLab's dataset of homomeric AlphaFold 2 models. These advancements significantly improve the functionality and accessibility of these resources, enabling discoveries using structure data.

• Waman et al. (2024) — Explains the CATH hierarchy (class, architecture, topology, homologous superfamily) and how it has been extended with predicted models. Illustrates how structure classification works in practice. (review; CC BY 4.0)

  CATH 2024: CATH-AlphaFlow Doubles the Number of Structures in CATH and Reveals Nearly 200 New Folds. Waman et al. (2024). Journal of Molecular Biology. doi:10.1016/j.jmb.2024.168551

  Abstract

  CATH (https://www.cathdb.info) classifies domain structures from experimental protein structures in the PDB and predicted structures in the AlphaFold Database (AFDB). To cope with the scale of the predicted data a new NextFlow workflow (CATH-AlphaFlow), has been developed to classify high-quality domains into CATH superfamilies and identify novel fold groups and superfamilies. CATH-AlphaFlow uses a novel state-of-the-art structure-based domain boundary prediction method (ChainSaw) for identifying domains in multi-domain proteins. We applied CATH-AlphaFlow to process PDB structures not classified in CATH and AFDB structures from 21 model organisms, expanding CATH by over 100%. Domains not classified in existing CATH superfamilies or fold groups were used to seed novel folds, giving 253 new folds from PDB structures (September 2023 release) and 96 from AFDB structures of proteomes of 21 model organisms. Where possible, functional annotations were obtained using (i) predictions from publicly available methods (ii) annotations from structural relatives in AFDB/UniProt50. We also predicted functional sites and highly conserved residues. Some folds are associated with important functions such as photosynthetic acclimation (in flowering plants), iron permease activity (in fungi) and post-natal spermatogenesis (in mice). CATH-AlphaFlow will allow us to identify many more CATH relatives in the AFDB, further characterising the protein structure landscape.

D.9.7 Protein structure prediction

• Bertoline et al. (2023) — Covers the history of structure prediction from homology modelling, threading and fragment assembly through coevolution to AlphaFold2. Follows the same order as the section and is a good overview for revision. (review; CC BY 4.0)

  Before and after AlphaFold2: An overview of protein structure prediction. Bertoline et al. (2023). Frontiers in Bioinformatics. doi:10.3389/fbinf.2023.1120370

  Abstract

  Three-dimensional protein structure is directly correlated with its function and its determination is critical to understanding biological processes and addressing human health and life science problems in general. Although new protein structures are experimentally obtained over time, there is still a large difference between the number of protein sequences placed in Uniprot and those with resolved tertiary structure. In this context, studies have emerged to predict protein structures by methods based on a template or free modeling. In the last years, different methods have been combined to overcome their individual limitations, until the emergence of AlphaFold2, which demonstrated that predicting protein structure with high accuracy at unprecedented scale is possible. Despite its current impact in the field, AlphaFold2 has limitations. Recently, new methods based on protein language models have promised to revolutionize the protein structural biology allowing the discovery of protein structure and function only from evolutionary patterns present on protein sequence. Even though these methods do not reach AlphaFold2 accuracy, they already covered some of its limitations, being able to predict with high accuracy more than 200 million proteins from metagenomic databases. In this mini-review, we provide an overview of the breakthroughs in protein structure prediction before and after AlphaFold2 emergence.

• Kovalevskiy et al. (2024) — Reviews how AlphaFold2 predictions have been validated against experiments and how they have been used. Written by DeepMind authors, so it should be read alongside independent assessments; not peer reviewed. (review; CC BY 4.0)

  AlphaFold two years on: validation and impact. Kovalevskiy et al. (2024). doi:10.48550/arXiv.2403.02124

  Abstract

  Two years on from the initial release of AlphaFold2 we have seen its widespread adoption as a structure prediction tool. Here we discuss some of the latest work based on AlphaFold2, with a particular focus on its use within the structural biology community. This encompasses use cases like speeding up structure determination itself, enabling new computational studies, and building new tools and workflows. We also look at the ongoing validation of AlphaFold2, as its predictions continue to be compared against large numbers of experimental structures to further delineate the model's capabilities and limitations.

• Rennie and Oliver (2025) — Discusses how to interpret predicted models and extends to complexes, ligands and other biomolecules, and proposes guidelines for reporting AlphaFold predictions. Relevant to the subsections on co-folding, limitations and checking a model. (review; CC BY 4.0)

  Emerging frontiers in protein structure prediction following the AlphaFold revolution. Rennie & Oliver (2025). Journal of The Royal Society Interface. doi:10.1098/rsif.2024.0886

  Abstract

  Models of protein structures enable molecular understanding of biological processes. Current protein structure prediction tools lie at the interface of biology, chemistry and computer science. Millions of protein structure models have been generated in a very short space of time through a revolution in protein structure prediction driven by deep learning, led by AlphaFold. This has provided a wealth of new structural information. Interpreting these predictions is critical to determining where and when this information is useful. But proteins are not static nor do they act alone, and structures of proteins interacting with other proteins and other biomolecules are critical to a complete understanding of their biological function at the molecular level. This review focuses on the application of state-of-the-art protein structure prediction to these advanced applications. We also suggest a set of guidelines for reporting AlphaFold predictions.

D.9.8 Protein language models and structure

• Weissenow and Rost (2025) — Discusses how embeddings from protein language models are replacing multiple sequence alignments as input for many prediction tasks, and the trade-offs in accuracy and cost. Directly relevant to the comparison of language models with alignments. (review; CC BY-NC-ND 4.0 (link only))

  Are protein language models the new universal key?. Weissenow & Rost (2025). Current Opinion in Structural Biology. doi:10.1016/j.sbi.2025.102997

  Abstract

  Protein language models (pLMs) capture some aspects of the grammar of the language of life as written in protein sequences. The so-called pLM embeddings implicitly contain this information. Therefore, embeddings can serve as the exclusive input into downstream supervised methods for protein prediction. Over the last 33 years, evolutionary information extracted through simple averaging for specific protein families from multiple sequence alignments (MSAs) has been the most successful universal key to the success of protein prediction. For many applications, MSA-free pLM-based predictions now have become significantly more accurate. The reason for this is often a combination of two aspects. Firstly, embeddings condense the grammar so efficiently that downstream prediction methods succeed with small models, i.e., they need few free parameters in particular in the era of exploding deep neural networks. Secondly, pLM-based methods provide protein-specific solutions. As additional benefit, once the pLM pre-training is complete, pLM-based solutions tend to consume much fewer resources than MSA-based solutions. In fact, we appeal to the community to rather optimize foundation models than to retrain new ones and to evolve incentives for solutions that require fewer resources even at some loss in accuracy. Although pLMs have not, yet, succeeded to entirely replace the body of solutions developed over three decades, they clearly are rapidly advancing as the universal key for protein prediction.

• Zhang et al. (2024) — A research article rather than a review: it analyses what ESM-2 has learned and argues that it mainly stores coevolution statistics of interacting motifs. Helps to understand why language-model structure prediction resembles alignment-based methods and where it fails. (research article; CC BY 4.0)

  Protein language models learn evolutionary statistics of interacting sequence motifs. Zhang et al. (2024). Proceedings of the National Academy of Sciences. doi:10.1073/pnas.2406285121

  Abstract

  Protein language models (pLMs) have emerged as potent tools for predicting and designing protein structure and function, and the degree to which these models fundamentally understand the inherent biophysics of protein structure stands as an open question. Motivated by a finding that pLM-based structure predictors erroneously predict nonphysical structures for protein isoforms, we investigated the nature of sequence context needed for contact predictions in the pLM Evolutionary Scale Modeling (ESM-2). We demonstrate by use of a "categorical Jacobian" calculation that ESM-2 stores statistics of coevolving residues, analogously to simpler modeling approaches like Markov Random Fields and Multivariate Gaussian models. We further investigated how ESM-2 "stores" information needed to predict contacts by comparing sequence masking strategies, and found that providing local windows of sequence information allowed ESM-2 to best recover predicted contacts. This suggests that pLMs predict contacts by storing motifs of pairwise contacts. Our investigation highlights the limitations of current pLMs and underscores the importance of understanding the underlying mechanisms of these models.

D.9.9 Membrane proteins

• Phillips and Miller (2021) — A short review of how α-helical membrane proteins are inserted and folded in the endoplasmic reticulum and how misfolded ones are recognised. Gives the biological context for topology and insertion. (review; CC BY 4.0)

  Membrane protein folding and quality control. Phillips & Miller (2021). Current Opinion in Structural Biology. doi:10.1016/j.sbi.2021.03.003

  Abstract

  Membrane proteins account for a quarter of cellular proteins, and most are synthesised at the endoplasmic reticulum (ER). Insertion and folding of polypeptides in the membrane environment is prone to error, necessitating diverse quality control systems. Recent discoveries have demonstrated how forces act on the nascent chain during insertion, and revealed new translocon components and accessories that facilitate the correct biogenesis of substrates. Our understanding of one of the best studied quality control systems-ER-associated degradation-has been advanced through new structural and functional studies of the core Hrd1 complex, and through the discovery of a new branch of this degradative pathway. New data also reveal how cells resolve clogged translocons, which would otherwise be unable to function. Finally, new work elucidates how mitochondrial tail-anchored proteins that have been mistargeted to the ER are identified and destroyed. Overall, we describe an emerging picture of an increasingly complex quality control network.

• Whitley et al. (2021) — Explains how the translocon recognises transmembrane helices, the hydrophobicity-based insertion scale and the role of charged residues in topology. Underpins the principles behind α-helical topology prediction. (review; CC BY 4.0)

  Folding and Insertion of Transmembrane Helices at the ER. Whitley et al. (2021). International Journal of Molecular Sciences. doi:10.3390/ijms222312778

  Abstract

  In eukaryotic cells, the endoplasmic reticulum (ER) is the entry point for newly synthesized proteins that are subsequently distributed to organelles of the endomembrane system. Some of these proteins are completely translocated into the lumen of the ER while others integrate stretches of amino acids into the greasy 30 Å wide interior of the ER membrane bilayer. It is generally accepted that to exist in this non-aqueous environment the majority of membrane integrated amino acids are primarily non-polar/hydrophobic and adopt an α-helical conformation. These stretches are typically around 20 amino acids long and are known as transmembrane (TM) helices. In this review, we will consider how transmembrane helices achieve membrane integration. We will address questions such as: Where do the stretches of amino acids fold into a helical conformation? What is/are the route/routes that these stretches take from synthesis at the ribosome to integration through the ER translocon? How do these stretches 'know' to integrate and in which orientation? How do marginally hydrophobic stretches of amino acids integrate and survive as transmembrane helices?

• Horne et al. (2020) — Reviews the structure, folding and membrane insertion of β-barrel outer membrane proteins and the role of the BAM complex and the lipid bilayer. Covers the β-barrel subsection. (review; CC BY 4.0)

  Role of the lipid bilayer in outer membrane protein folding in Gram-negative bacteria. Horne et al. (2020). Journal of Biological Chemistry. doi:10.1074/jbc.rev120.011473

  Abstract

  β-Barrel outer membrane proteins (OMPs) represent the major proteinaceous component of the outer membrane (OM) of Gram-negative bacteria. These proteins perform key roles in cell structure and morphology, nutrient acquisition, colonization and invasion, and protection against external toxic threats such as antibiotics. To become functional, OMPs must fold and insert into a crowded and asymmetric OM that lacks much freely accessible lipid. This feat is accomplished in the absence of an external energy source and is thought to be driven by the high thermodynamic stability of folded OMPs in the OM. With such a stable fold, the challenge that bacteria face in assembling OMPs into the OM is how to overcome the initial energy barrier of membrane insertion. In this review, we highlight the roles of the lipid environment and the OM in modulating the OMP-folding landscape and discuss the factors that guide folding in vitro and in vivo We particularly focus on the composition, architecture, and physical properties of the OM and how an understanding of the folding properties of OMPs in vitro can help explain the challenges they encounter during folding in vivo Current models of OMP biogenesis in the cellular environment are still in flux, but the stakes for improving the accuracy of these models are high. OMP folding is an essential process in all Gram-negative bacteria, and considering the looming crisis of widespread microbial drug resistance it is an attractive target. To bring down this vital OMP-supported barrier to antibiotics, we must first understand how bacterial cells build it.

D.9.10 Dynamics and docking

• Braun et al. (2018) — Introduces the foundations of molecular simulation: force fields, integration, ensembles, sampling and statistical uncertainty. Written for beginners and useful for understanding what molecular dynamics can and cannot tell you. (tutorial; CC BY 4.0)

  Best Practices for Foundations in Molecular Simulations [Article v1.0]. Braun et al. (2018). Living Journal of Computational Molecular Science. doi:10.33011/livecoms.1.1.5957

  Abstract

  This document provides a starting point for approaching molecular simulations, guiding beginning practitioners to what issues they need to know about before and while starting their first simulations, and why those issues are so critical. This document makes no claims to provide an adequate introduction to the subject on its own. Instead, our goal is to help people know what issues are critical before beginning, and to provide references to good resources on those topics. We also provide a checklist of key issues to consider before and while setting up molecular simulations which may serve as a foundation for other best practices documents.

• Togashi and Flechsig (2018) — Explains elastic network models and normal-mode analysis and how they are used to describe large-scale protein motions. Covers the subsection on normal modes. (review; CC BY 4.0)

  Coarse-Grained Protein Dynamics Studies Using Elastic Network Models. Togashi & Flechsig (2018). International Journal of Molecular Sciences. doi:10.3390/ijms19123899

  Abstract

  Elastic networks have been used as simple models of proteins to study their slow structural dynamics. They consist of point-like particles connected by linear Hookean springs and hence are convenient for linear normal mode analysis around a given reference structure. Furthermore, dynamic simulations using these models can provide new insights. As the computational cost associated with these models is considerably lower compared to that of all-atom models, they are also convenient for comparative studies between multiple protein structures. In this review, we introduce examples of coarse-grained molecular dynamics studies using elastic network models and their derivatives, focusing on the nonlinear phenomena, and discuss their applicability to large-scale macromolecular assemblies.

• Paggi et al. (2024) — A clear review of how docking works (sampling and scoring), why it succeeds or fails, and how deep-learning methods compare with physics-based docking. Directly relevant to the subsection on docking and scoring functions. (review; CC BY 4.0)

  The Art and Science of Molecular Docking. Paggi et al. (2024). Annual Review of Biochemistry. doi:10.1146/annurev-biochem-030222-120000

  Abstract

  Molecular docking has become an essential part of a structural biologist's and medicinal chemist's toolkits. Given a chemical compound and the three-dimensional structure of a molecular target-for example, a protein-docking methods fit the compound into the target, predicting the compound's bound structure and binding energy. Docking can be used to discover novel ligands for a target by screening large virtual compound libraries. Docking can also provide a useful starting point for structure-based ligand optimization or for investigating a ligand's mechanism of action. Advances in computational methods, including both physics-based and machine learning approaches, as well as in complementary experimental techniques, are making docking an even more powerful tool. We review how docking works and how it can drive drug discovery and biological research. We also describe its current limitations and ongoing efforts to overcome them.

D.9.11 Protein design

• Pan and Kortemme (2021) — Explains the principles of physics-based de novo design, including energy functions and sequence and backbone design, and the early deep-learning approaches. Good for the physics-based design subsection. (review; CC BY 4.0)

  Recent advances in de novo protein design: Principles, methods, and applications. Pan & Kortemme (2021). Journal of Biological Chemistry. doi:10.1016/j.jbc.2021.100558

  Abstract

  The computational de novo protein design is increasingly applied to address a number of key challenges in biomedicine and biological engineering. Successes in expanding applications are driven by advances in design principles and methods over several decades. Here, we review recent innovations in major aspects of the de novo protein design and include how these advances were informed by principles of protein architecture and interactions derived from the wealth of structures in the Protein Data Bank. We describe developments in de novo generation of designable backbone structures, optimization of sequences, design scoring functions, and the design of the function. The advances not only highlight design goals reachable now but also point to the challenges and opportunities for the future of the field.

• Winnifrith et al. (2024) — A short overview of generative models for protein design, including diffusion models for backbones, inverse folding and language models for sequences. Covers the deep-learning design subsection. (review; CC BY 4.0)

  Generative artificial intelligence for de novo protein design. Winnifrith et al. (2024). Current Opinion in Structural Biology. doi:10.1016/j.sbi.2024.102794

  Abstract

  Engineering new molecules with desirable functions and properties has the potential to extend our ability to engineer proteins beyond what nature has so far evolved. Advances in the so-called 'de novo' design problem have recently been brought forward by developments in artificial intelligence. Generative architectures, such as language models and diffusion processes, seem adept at generating novel, yet realistic proteins that display desirable properties and perform specified functions. State-of-the-art design protocols now achieve experimental success rates nearing 20%, thus widening the access to de novo designed proteins. Despite extensive progress, there are clear field-wide challenges, for example, in determining the best in silico metrics to prioritise designs for experimental testing, and in designing proteins that can undergo large conformational changes or be regulated by post-translational modifications. With an increase in the number of models being developed, this review provides a framework to understand how these tools fit into the overall process of de novo protein design. Throughout, we highlight the power of incorporating biochemical knowledge to improve performance and interpretability.

• Jänes and Beltrao (2024) — Reviews deep-learning structure prediction beyond single chains (complexes, alternative conformations, evolution) and its use in protein design. Links the design section back to the prediction methods it builds on. (review; CC BY 4.0)

  Deep learning for protein structure prediction and design—progress and applications. Jänes & Beltrao (2024). Molecular Systems Biology. doi:10.1038/s44320-024-00016-x

  Abstract

  Proteins are the key molecular machines that orchestrate all biological processes of the cell. Most proteins fold into three-dimensional shapes that are critical for their function. Studying the 3D shape of proteins can inform us of the mechanisms that underlie biological processes in living cells and can have practical applications in the study of disease mutations or the discovery of novel drug treatments. Here, we review the progress made in sequence-based prediction of protein structures with a focus on applications that go beyond the prediction of single monomer structures. This includes the application of deep learning methods for the prediction of structures of protein complexes, different conformations, the evolution of protein structures and the application of these methods to protein design. These developments create new opportunities for research that will have impact across many areas of biomedical research.

D.10 Protein function

Further explanations for Chapter 10.

D.10.1 Definitions of function

• Thomas (2016) — Discusses what ‘function’ means for a gene product and how the GO distinguishes molecular function, biological process and cellular component. Useful background for the section’s opening question of how to define function. (book chapter; CC BY 4.0)

  The Gene Ontology and the Meaning of Biological Function. Thomas (2016). The Gene Ontology Handbook. doi:10.1007/978-1-4939-3743-1_2

  Abstract

  The Gene Ontology (GO) provides a framework and set of concepts for describing the functions of gene products from all organisms. It is specifically designed for supporting the computational representation of biological systems. A GO annotation is an association between a specific gene product and a GO concept, together making a statement pertinent to the function of that gene. However, the meaning of the term "function" is not as straightforward as it might seem, and has been discussed at length in both philosophical and biological circles. Here, I first review these discussions. I then present an explicit formulation of the biological model that underlies the GO and annotations, and discuss how this model relates to the broader debates on the meaning of biological function.

• Gaudet et al. (2016) — A short introduction to the structure of the GO, its relations, annotations and evidence codes. A practical companion to the subsection on the Gene Ontology. (primer; CC BY 4.0)

  Primer on the Gene Ontology. Gaudet et al. (2016). The Gene Ontology Handbook. doi:10.1007/978-1-4939-3743-1_3

  Abstract

  The Gene Ontology (GO) project is the largest resource for cataloguing gene function. The combination of solid conceptual underpinnings and a practical set of features have made the GO a widely adopted resource in the research community and an essential resource for data analysis. In this chapter, we provide a concise primer for all users of the GO. We briefly introduce the structure of the ontology and explain how to interpret annotations associated with the GO.

D.10.2 Function prediction

• Vu and Jung (2021) — Reviews GO-based function prediction from homology, domain and network methods to deep learning, including evaluation. Covers most of the subsections of this section in one place. (review; CC BY 4.0)

  Protein function prediction with gene ontology: from traditional to deep learning models. Vu & Jung (2021). PeerJ. doi:10.7717/peerj.12019

  Abstract

  Protein function prediction is a crucial part of genome annotation. Prediction methods have recently witnessed rapid development, owing to the emergence of high-throughput sequencing technologies. Among the available databases for identifying protein function terms, Gene Ontology (GO) is an important resource that describes the functional properties of proteins. Researchers are employing various approaches to efficiently predict the GO terms. Meanwhile, deep learning, a fast-evolving discipline in data-driven approach, exhibits impressive potential with respect to assigning GO terms to amino acid sequences. Herein, we reviewed the currently available computational GO annotation methods for proteins, ranging from conventional to deep learning approach. Further, we selected some suitable predictors from among the reviewed tools and conducted a mini comparison of their performance using a worldwide challenge dataset. Finally, we discussed the remaining major challenges in the field, and emphasized the future directions for protein function prediction with GO.

• Cozzetto and Jones (2016) — Explains annotation transfer by sequence similarity, orthology and domain/family signatures, and the errors this can propagate. Useful for the subsections on homology- and family-based annotation. (book chapter; CC BY 4.0)

  Computational Methods for Annotation Transfers from Sequence. Cozzetto & Jones (2016). The Gene Ontology Handbook. doi:10.1007/978-1-4939-3743-1_5

  Abstract

  Surveys of public sequence resources show that experimentally supported functional information is still completely missing for a considerable fraction of known proteins and is clearly incomplete for an even larger portion. Bioinformatics methods have long made use of very diverse data sources alone or in combination to predict protein function, with the understanding that different data types help elucidate complementary biological roles. This chapter focuses on methods accepting amino acid sequences as input and producing GO term assignments directly as outputs; the relevant biological and computational concepts are presented along with the advantages and limitations of individual approaches.

• Lin et al. (2024) — A recent survey grouping function predictors by the information they use (sequence, structure, networks, language models) and comparing them. Gives a more up-to-date view of the machine-learning methods. (review; CC BY 4.0)

  A comprehensive review and comparison of existing computational methods for protein function prediction. Lin et al. (2024). Briefings in Bioinformatics. doi:10.1093/bib/bbae289

  Abstract

  Protein function prediction is critical for understanding the cellular physiological and biochemical processes, and it opens up new possibilities for advancements in fields such as disease research and drug discovery. During the past decades, with the exponential growth of protein sequence data, many computational methods for predicting protein function have been proposed. Therefore, a systematic review and comparison of these methods are necessary. In this study, we divide these methods into four different categories, including sequence-based methods, 3D structure-based methods, PPI network-based methods and hybrid information-based methods. Furthermore, their advantages and disadvantages are discussed, and then their performance is comprehensively evaluated and compared. Finally, we discuss the challenges and opportunities present in this field.

D.10.3 Subcellular localisation

• Imai and Nakai (2020) — Describes the main sorting signals (signal peptides, transit peptides, nuclear signals) and the tools that recognise them, written by authors of PSORT. Closely matches the section’s structure from signals to prediction. (review; CC BY 4.0)

  Tools for the Recognition of Sorting Signals and the Prediction of Subcellular Localization of Proteins From Their Amino Acid Sequences. Imai & Nakai (2020). Frontiers in Genetics. doi:10.3389/fgene.2020.607812

  Abstract

  At the time of translation, nascent proteins are thought to be sorted into their final subcellular localization sites, based on the part of their amino acid sequences (i.e., sorting or targeting signals). Thus, it is interesting to computationally recognize these signals from the amino acid sequences of any given proteins and to predict their final subcellular localization with such information, supplemented with additional information (e.g., k -mer frequency). This field has a long history and many prediction tools have been released. Even in this era of proteomic atlas at the single-cell level, researchers continue to develop new algorithms, aiming at accessing the impact of disease-causing mutations/cell type-specific alternative splicing, for example. In this article, we overview the entire field and discuss its future direction.

• Gillani and Pollastri (2024) — A recent overview of localisation predictors, including deep-learning and language-model-based methods, and the datasets used to train them. (review; CC BY 4.0)

  Protein subcellular localization prediction tools <sup/>. Gillani & Pollastri (2024). Computational and Structural Biotechnology Journal. doi:10.1016/j.csbj.2024.04.032

  Abstract

  Protein subcellular localization prediction is of great significance in bioinformatics and biological research. Most of the proteins do not have experimentally determined localization information, computational prediction methods and tools have been acting as an active research area for more than two decades now. Knowledge of the subcellular location of a protein provides valuable information about its functionalities, the functioning of the cell, and other possible interactions with proteins. Fast, reliable, and accurate predictors provides platforms to harness the abundance of sequence data to predict subcellular locations accordingly. During the last decade, there has been a considerable amount of research effort aimed at developing subcellular localization predictors. This paper reviews recent subcellular localization prediction tools in the Eukaryotic, Prokaryotic, and Virus-based categories followed by a detailed analysis. Each predictor is discussed based on its main features, strengths, weaknesses, algorithms used, prediction techniques, and analysis. This review is supported by prediction tools taxonomies that highlight their rele- vant area and examples for uncomplicated categorization and ease of understandability. These taxonomies help users find suitable tools according to their needs. Furthermore, recent research gaps and challenges are discussed to cover areas that need the utmost attention. This survey provides an in-depth analysis of the most recent prediction tools to facilitate readers and can be considered a quick guide for researchers to identify and explore the recent literature advancements.

D.10.4 Predicting the effects of mutations

• Riccio et al. (2024) — A short, practical overview of variant effect predictors, how they are built and how to choose and interpret them. Relevant to the subsection on using predictions. (review; CC BY 4.0)

  Variant effect predictors: a systematic review and practical guide. Riccio et al. (2024). Human Genetics. doi:10.1007/s00439-024-02670-5

  Abstract

  Large-scale association analyses using whole-genome sequence data have become feasible, but understanding the functional impacts of these associations remains challenging. Although many tools are available to predict the functional impacts of genetic variants, it is unclear which tool should be used in practice. This work provides a practical guide to assist in selecting appropriate tools for variant annotation. We conducted a MEDLINE search up to November 10, 2023, and included tools that are applicable to a broad range of phenotypes, can be used locally, and have been recently updated. Tools were categorized based on the types of variants they accept and the functional impacts they predict. Sequence Ontology terms were used for standardization. We identified 118 databases and software packages, encompassing 36 variant types and 161 functional impacts. Combining only three tools, namely SnpEff, FAVOR, and SparkINFERNO, allows predicting 99 (61%) distinct functional impacts. Thirty-seven tools predict 89 functional impacts that are not supported by any other tool, while 75 tools predict pathogenicity and can be used within the ACMG/AMP guidelines in a clinical context. We launched a website allowing researchers to select tools based on desired variants and impacts. In summary, more than 100 tools are already available to predict approximately 160 functional impacts. About 60% of the functional impacts can be predicted by the combination of three tools. Unexpectedly, recent tools do not predict more impacts than older ones. Future research should allow predicting the functionality of so far unsupported variant types, such as gene fusions.URL: https://cardio-care.shinyapps.io/VEP_Finder/ .Registration: OSF Registries on November 10, 2023, https://osf.io/s2gct .

• Livesey and Marsh (2023) — Benchmarks many predictors, including unsupervised evolutionary models, against deep mutational scanning data. A primary benchmark rather than a tutorial, but it illustrates the experimental-benchmark subsection well. (review; CC BY 4.0)

  Updated benchmarking of variant effect predictors using deep mutational scanning. Livesey & Marsh (2023). Molecular Systems Biology. doi:10.15252/msb.202211474

  Abstract

  The assessment of variant effect predictor (VEP) performance is fraught with biases introduced by benchmarking against clinical observations. In this study, building on our previous work, we use independently generated measurements of protein function from deep mutational scanning (DMS) experiments for 26 human proteins to benchmark 55 different VEPs, while introducing minimal data circularity. Many top-performing VEPs are unsupervised methods including EVE, DeepSequence and ESM-1v, a protein language model that ranked first overall. However, the strong performance of recent supervised VEPs, in particular VARITY, shows that developers are taking data circularity and bias issues seriously. We also assess the performance of DMS and unsupervised VEPs for discriminating between known pathogenic and putatively benign missense variants. Our findings are mixed, demonstrating that some DMS datasets perform exceptionally at variant classification, while others are poor. Notably, we observe a striking correlation between VEP agreement with DMS data and performance in identifying clinically relevant variants, strongly supporting the validity of our rankings and the utility of DMS for independent benchmarking.

• Gerasimavicius et al. (2025) — A brief review of how protein structure, including predicted structures, can be combined with sequence-based predictors to assess missense variants. (review; CC BY 4.0)

  Leveraging protein structural information to improve variant effect prediction. Gerasimavicius et al. (2025). Current Opinion in Structural Biology. doi:10.1016/j.sbi.2025.103023

  Abstract

  Despite massive sequencing efforts, understanding the difference between human pathogenic and benign variants remains a challenge. Computational variant effect predictors (VEPs) have emerged as essential tools for assessing the impact of genetic variants, although their performance varies. Initially, sequence-based methods dominated the field, but recent advances, particularly in protein structure prediction technologies like AlphaFold, have led to an increased utilization of structural information by VEPs aimed at scoring human missense variants. This review highlights the progress in integrating structural information into VEPs, showcasing novel models such as AlphaMissense, PrimateAI-3D, and CPT-1 that demonstrate improved variant evaluation. Structural data offers more interpretability, especially for non-loss-of-function variants, and provides insights into complex variant interactions in vivo. As the field advances, utilizing biomolecular complex structures will be pivotal for future VEP development, with recent breakthroughs in protein-ligand and protein-nucleic acid complex prediction offering new avenues.

D.10.5 Benchmarking function prediction

• Friedberg and Radivojac (2016) — Written by CAFA organisers; explains the design of CAFA, its metrics and the difficulties of evaluating function prediction with incomplete annotations. (book chapter; CC BY 4.0)

  Community-Wide Evaluation of Computational Function Prediction. Friedberg & Radivojac (2016). The Gene Ontology Handbook. doi:10.1007/978-1-4939-3743-1_10

  Abstract

  A biological experiment is the most reliable way of assigning function to a protein. However, in the era of high-throughput sequencing, scientists are unable to carry out experiments to determine the function of every single gene product. Therefore, to gain insights into the activity of these molecules and guide experiments, we must rely on computational means to functionally annotate the majority of sequence data. To understand how well these algorithms perform, we have established a challenge involving a broad scientific community in which we evaluate different annotation methods according to their ability to predict the associations between previously unannotated protein sequences and Gene Ontology terms. Here we discuss the rationale, benefits, and issues associated with evaluating computational methods in an ongoing community-wide challenge.

• Škunca et al. (2016) — Discusses why assessing GO predictions is hard (open-world annotations, hierarchy, annotation biases) and how evaluation strategies address this. Matches the subsection on why function prediction is hard to evaluate. (book chapter; CC BY 4.0)

  Evaluating Computational Gene Ontology Annotations. Škunca et al. (2016). The Gene Ontology Handbook. doi:10.1007/978-1-4939-3743-1_8

  Abstract

  Two avenues to understanding gene function are complementary and often overlapping: experimental work and computational prediction. While experimental annotation generally produces high-quality annotations, it is low throughput. Conversely, computational annotations have broad coverage, but the quality of annotations may be variable, and therefore evaluating the quality of computational annotations is a critical concern.In this chapter, we provide an overview of strategies to evaluate the quality of computational annotations. First, we discuss why evaluating quality in this setting is not trivial. We highlight the various issues that threaten to bias the evaluation of computational annotations, most of which stem from the incompleteness of biological databases. Second, we discuss solutions that address these issues, for example, targeted selection of new experimental annotations and leveraging the existing experimental annotations.

D.10.6 Experimental determination of function

• Bock et al. (2022) — A primer on pooled and arrayed CRISPR screens, their readouts and analysis; linked only. (primer; not open access)

  High-content CRISPR screening. Bock et al. (2022). Nature Reviews Methods Primers. doi:10.1038/s43586-021-00093-4

D.11 Protein–protein interactions

Further explanations for Chapter 11.

D.11.1 Experimental methods

• Snider et al. (2015) — Explains the principles of interactome mapping, compares binary assays (yeast two-hybrid and relatives) with co-complex methods such as AP-MS, and discusses their strengths, weaknesses, follow-up validation and the main interaction databases. A good companion to the whole section, including errors, biases and coverage. (review; CC BY 4.0)

  Fundamentals of protein interaction network mapping. Snider et al. (2015). Molecular Systems Biology. doi:10.15252/msb.20156351

  Abstract

  Studying protein interaction networks of all proteins in an organism ("interactomes") remains one of the major challenges in modern biomedicine. Such information is crucial to understanding cellular pathways and developing effective therapies for the treatment of human diseases. Over the past two decades, diverse biochemical, genetic, and cell biological methods have been developed to map interactomes. In this review, we highlight basic principles of interactome mapping. Specifically, we discuss the strengths and weaknesses of individual assays, how to select a method appropriate for the problem being studied, and provide general guidelines for carrying out the necessary follow-up analyses. In addition, we discuss computational methods to predict, map, and visualize interactomes, and provide a summary of some of the most important interactome resources. We hope that this review serves as both a useful overview of the field and a guide to help more scientists actively employ these powerful approaches in their research.

• Adames et al. (2018) — Reviews how multi-mutant yeast strains are made for genetic interaction screens, comparing classical methods with CRISPR/Cas-based approaches. Useful technical background to the large yeast genetic interaction maps discussed in the subsection on genetic interactions. (review; CC BY 4.0)

  Yeast genetic interaction screens in the age of CRISPR/Cas. Adames et al. (2018). Current Genetics. doi:10.1007/s00294-018-0887-8

  Abstract

  The ease of performing both forward and reverse genetics in Saccharomyces cerevisiae, along with its stable haploid state and short generation times, has made this budding yeast the consummate model eukaryote for genetics. The major advantage of using budding yeast for reverse genetics is this organism's highly efficient homology-directed repair, allowing for precise genome editing simply by introducing DNA with homology to the chromosomal target. Although plasmid- and PCR-based genome editing tools are quite efficient, they depend on rare spontaneous DNA breaks near the target sequence. Consequently, they can generate only one genomic edit at a time, and the edit must be associated with a selectable marker. However, CRISPR/Cas technology is efficient enough to permit markerless and multiplexed edits in a single step. These features have made CRISPR/Cas popular for yeast strain engineering in synthetic biology and metabolic engineering applications, but it has not been widely employed for genetic screens. In this review, we critically examine different methods to generate multi-mutant strains in systematic genetic interaction screens and discuss the potential of CRISPR/Cas to supplement or improve on these methods.

• Marku and Pancaldi (2023) — An overview, written for newcomers, of the main types of algorithms for inferring gene regulatory networks from time-series transcriptomics, with guidance on choosing a method. Useful background for the subsection on network inference. (review; CC BY 4.0)

  From time-series transcriptomics to gene regulatory networks: A review on inference methods. Marku & Pancaldi (2023). PLOS Computational Biology. doi:10.1371/journal.pcbi.1011254

  Abstract

  Inference of gene regulatory networks has been an active area of research for around 20 years, leading to the development of sophisticated inference algorithms based on a variety of assumptions and approaches. With the ever increasing demand for more accurate and powerful models, the inference problem remains of broad scientific interest. The abstract representation of biological systems through gene regulatory networks represents a powerful method to study such systems, encoding different amounts and types of information. In this review, we summarize the different types of inference algorithms specifically based on time-series transcriptomics, giving an overview of the main applications of gene regulatory networks in computational biology. This review is intended to give an updated reference of regulatory networks inference tools to biologists and researchers new to the topic and guide them in selecting the appropriate inference method that best fits their questions, aims, and experimental data.

D.11.2 Networks

• Koutrouli et al. (2020) — An introduction to graph theory for biologists: graph types, degree distributions, clustering, motifs, network models, clustering algorithms and tools. Covers the network concepts used throughout the section. (review; CC BY 4.0)

  A Guide to Conquer the Biological Network Era Using Graph Theory. Koutrouli et al. (2020). Frontiers in Bioengineering and Biotechnology. doi:10.3389/fbioe.2020.00034

  Abstract

  Networks are one of the most common ways to represent biological systems as complex sets of binary interactions or relations between different bioentities. In this article, we discuss the basic graph theory concepts and the various graph types, as well as the available data structures for storing and reading graphs. In addition, we describe several network properties and we highlight some of the widely used network topological features. We briefly mention the network patterns, motifs and models, and we further comment on the types of biological and biomedical networks along with their corresponding computer- and human-readable file formats. Finally, we discuss a variety of algorithms and metrics for network analyses regarding graph drawing, clustering, visualization, link prediction, perturbation, and network alignment as well as the current state-of-the-art tools. We expect this review to reach a very broad spectrum of readers varying from experts to beginners while encouraging them to enhance the field further.

• Sonawane et al. (2019) — Surveys network types and data sources used in network medicine, and how protein interaction, co-expression and gene regulatory networks are combined with phenotype-specific data to study disease, with example applications. (review; CC BY 4.0)

  Network Medicine in the Age of Biomedical Big Data. Sonawane et al. (2019). Frontiers in Genetics. doi:10.3389/fgene.2019.00294

  Abstract

  Network medicine is an emerging area of research dealing with molecular and genetic interactions, network biomarkers of disease, and therapeutic target discovery. Large-scale biomedical data generation offers a unique opportunity to assess the effect and impact of cellular heterogeneity and environmental perturbations on the observed phenotype. Marrying the two, network medicine with biomedical data provides a framework to build meaningful models and extract impactful results at a network level. In this review, we survey existing network types and biomedical data sources. More importantly, we delve into ways in which the network medicine approach, aided by phenotype-specific biomedical data, can be gainfully applied. We provide three paradigms, mainly dealing with three major biological network archetypes: protein-protein interaction, expression-based, and gene regulatory networks. For each of these paradigms, we discuss a broad overview of philosophies under which various network methods work. We also provide a few examples in each paradigm as a test case of its successful application. Finally, we delineate several opportunities and challenges in the field of network medicine. We hope this review provides a lexicon for researchers from biological sciences and network theory to come on the same page to work on research areas that require interdisciplinary expertise. Taken together, the understanding gained from combining biomedical data with networks can be useful for characterizing disease etiologies and identifying therapeutic targets, which, in turn, will lead to better preventive medicine with translational impact on personalized healthcare.

• Gu et al. (2019) — Reviews genome-scale metabolic models, which are simulated to predict metabolic fluxes, and their applications from strain engineering to drug targets and human disease; background for the subsection on systems biology and metabolic modelling. (review; CC BY 4.0)

  Current status and applications of genome-scale metabolic models. Gu et al. (2019). Genome Biology. doi:10.1186/s13059-019-1730-3

  Abstract

  Genome-scale metabolic models (GEMs) computationally describe gene-protein-reaction associations for entire metabolic genes in an organism, and can be simulated to predict metabolic fluxes for various systems-level metabolic studies. Since the first GEM for Haemophilus influenzae was reported in 1999, advances have been made to develop and simulate GEMs for an increasing number of organisms across bacteria, archaea, and eukarya. Here, we review current reconstructed GEMs and discuss their applications, including strain development for chemicals and materials production, drug targeting in pathogens, prediction of enzyme functions, pan-reactome analysis, modeling interactions among multiple cells or organisms, and understanding human diseases.

D.11.3 Prediction methods

• Tsuchiya et al. (2022) — A short review that traces computational prediction of protein complexes from rigid-body docking and scoring to AlphaFold-based methods, with a historical perspective. Useful for the docking and deep learning subsections. (review; CC BY 4.0)

  Protein–protein interaction prediction methods: from docking-based to AI-based approaches. Tsuchiya et al. (2022). Biophysical Reviews. doi:10.1007/s12551-022-01032-7

  Abstract

  Protein-protein interactions (PPIs), such as protein-protein inhibitor, antibody-antigen complex, and supercomplexes play diverse and important roles in cells. Recent advances in structural analysis methods, including cryo-EM, for the determination of protein complex structures are remarkable. Nevertheless, much room remains for improvement and utilization of computational methods to predict PPIs because of the large number and great diversity of unresolved complex structures. This review introduces a wide array of computational methods, including our own, for estimating PPIs including antibody-antigen interactions, offering both historical and forward-looking perspectives.

• Kiouri et al. (2025) — Surveys machine-learning and deep-learning methods that use protein structure to predict interactions and interface residues, and discusses limitations such as scarce structural data and the choice of negative examples, which relates to the evaluation subsection. (review; CC BY 4.0)

  Structure-Based Approaches for Protein–Protein Interaction Prediction Using Machine Learning and Deep Learning. Kiouri et al. (2025). Biomolecules. doi:10.3390/biom15010141

  Abstract

  Protein-Protein Interaction (PPI) prediction plays a pivotal role in understanding cellular processes and uncovering molecular mechanisms underlying health and disease. Structure-based PPI prediction has emerged as a robust alternative to sequence-based methods, offering greater biological accuracy by integrating three-dimensional spatial and biochemical features. This work summarizes the recent advances in computational approaches leveraging protein structure information for PPI prediction, focusing on machine learning (ML) and deep learning (DL) techniques. These methods not only improve predictive accuracy but also provide insights into functional sites, such as binding and catalytic residues. However, challenges such as limited high-resolution structural data and the need for effective negative sampling persist. Through the integration of experimental and computational tools, structure-based prediction paves the way for comprehensive proteomic network analysis, holding promise for advancements in drug discovery, biomarker identification, and personalized medicine. Future directions include enhancing scalability and dataset reliability to expand these approaches across diverse proteomes.

• Lin et al. (2025) — Reviews the use of AlphaFold 3 for complex structure prediction, compares it with docking tools such as ClusPro and with AlphaPulldown, and discusses its limitations for large complexes, dynamics and mutation effects. The examples are from plant and fungal research but the discussion is general. (review; CC BY 4.0)

  Analysing protein complexes in plant science: insights and limitation with AlphaFold 3. Lin et al. (2025). Botanical Studies. doi:10.1186/s40529-025-00462-2

  Abstract

  AlphaFold 3 (AF3), an artificial intelligence (AI)-based software for protein complex structure prediction, represents a significant advancement in structural biology. Its flexibility and enhanced scalability have unlocked new applications in various fields, specifically in plant science, including improving crop resilience and predicting the structures of plant-specific proteins involved in stress responses, signalling pathways, and immune responses. Comparisons with existing tools, such as ClusPro and AlphaPulldown, highlight AF3's unique strengths in sequence-based interaction predictions and its greater adaptability to various biomolecular structures. However, limitations persist, including challenges in modelling large complexes, protein dynamics, and structures from underrepresented plant proteins with limited evolutionary data. Additionally, AF3 encounters difficulties in predicting mutation effects on protein interactions and DNA binding, which can be improved with molecular dynamics and experimental validation. This review presents an overview of AF3's advancements, using examples in plant and fungal research, and comparisons with existing tools. It also discusses current limitations and offers perspectives on integrating molecular dynamics and experimental validation to enhance its capabilities.

D.12 Machine learning

Further explanations for Chapter 12.

D.12.1 Basics

• Chicco (2017) — Practical advice on the whole workflow, from data preparation and splitting to hyperparameter tuning, imbalanced classes and the choice of evaluation metric. A short and accessible complement to the section on generalisation, overfitting and evaluation. (tutorial; CC BY 4.0)

  Ten quick tips for machine learning in computational biology. Chicco (2017). BioData Mining. doi:10.1186/s13040-017-0155-3

  Abstract

  Machine learning has become a pivotal tool for many projects in computational biology, bioinformatics, and health informatics. Nevertheless, beginners and biomedical researchers often do not have enough experience to run a data mining project effectively, and therefore can follow incorrect practices, that may lead to common mistakes or over-optimistic results. With this review, we present ten quick tips to take advantage of machine learning in any computational biology context, by avoiding some common errors that we observed hundreds of times in multiple bioinformatics projects. We believe our ten suggestions can strongly help any machine learning practitioner to carry on a successful project in computational biology and related sciences.

• Miller et al. (2024) — Overviews evaluation metrics for clustering, classification and regression, their advantages and disadvantages, and common pitfalls in model evaluation, with examples from genomics. Useful for the evaluation subsection. (review; CC BY 4.0)

  A review of model evaluation metrics for machine learning in genetics and genomics. Miller et al. (2024). Frontiers in Bioinformatics. doi:10.3389/fbinf.2024.1457619

  Abstract

  Machine learning (ML) has shown great promise in genetics and genomics where large and complex datasets have the potential to provide insight into many aspects of disease risk, pathogenesis of genetic disorders, and prediction of health and wellbeing. However, with this possibility there is a responsibility to exercise caution against biases and inflation of results that can have harmful unintended impacts. Therefore, researchers must understand the metrics used to evaluate ML models which can influence the critical interpretation of results. In this review we provide an overview of ML metrics for clustering, classification, and regression and highlight the advantages and disadvantages of each. We also detail common pitfalls that occur during model evaluation. Finally, we provide examples of how researchers can assess and utilise the results of ML models, specifically from a genomics perspective.

D.12.2 A short history of machine learning in bioinformatics

• Ben-Hur et al. (2008) — A tutorial introduction to support vector machines and kernels, including kernels for biological sequences, written for computational biologists. It explains the margin, soft margins and model selection clearly. (tutorial; CC BY 4.0)

  Support Vector Machines and Kernels for Computational Biology. Ben-Hur et al. (2008). PLoS Computational Biology. doi:10.1371/journal.pcbi.1000173

• Yoon (2009) — Introduces hidden Markov models, the Viterbi, forward and Baum–Welch algorithms, and profile HMMs, with applications in sequence analysis. Gives more detail than the section on how HMMs are decoded and trained. (review; CC BY (Europe PMC))

  Hidden Markov Models and their Applications in Biological Sequence Analysis. Yoon (2009). Current Genomics. doi:10.2174/138920209789177575

  Abstract

  Hidden Markov models (HMMs) have been extensively used in biological sequence analysis. In this paper, we give a tutorial review of HMMs and their applications in a variety of problems in molecular biology. We especially focus on three types of HMMs: the profile-HMMs, pair-HMMs, and context-sensitive HMMs. We show how these HMMs can be used to solve various sequence analysis problems, such as pairwise and multiple sequence alignments, gene annotation, classification, similarity search, and many others.

• Touw et al. (2012) — Describes how random forests work and how they are used in the life sciences, including variable importance measures and their pitfalls. Licence is non-commercial, so it is suitable only as a link. (review; CC BY-NC (link only))

  Data mining in the Life Sciences with Random Forest: a walk in the park or lost in the jungle?. Touw et al. (2012). Briefings in Bioinformatics. doi:10.1093/bib/bbs034

  Abstract

  In the Life Sciences 'omics' data is increasingly generated by different high-throughput technologies. Often only the integration of these data allows uncovering biological insights that can be experimentally validated or mechanistically modelled, i.e. sophisticated computational approaches are required to extract the complex non-linear trends present in omics data. Classification techniques allow training a model based on variables (e.g. SNPs in genetic association studies) to separate different classes (e.g. healthy subjects versus patients). Random Forest (RF) is a versatile classification algorithm suited for the analysis of these large data sets. In the Life Sciences, RF is popular because RF classification models have a high-prediction accuracy and provide information on importance of variables for classification. For omics data, variables or conditional relations between variables are typically important for a subset of samples of the same class. For example: within a class of cancer patients certain SNP combinations may be important for a subset of patients that have a specific subtype of cancer, but not important for a different subset of patients. These conditional relationships can in principle be uncovered from the data with RF as these are implicitly taken into account by the algorithm during the creation of the classification model. This review details some of the to the best of our knowledge rarely or never used RF properties that allow maximizing the biological insights that can be extracted from complex omics data sets using RF.

D.12.3 Deep learning

• Angermueller et al. (2016) — An introduction to deep neural networks for biologists, covering training, regularisation, convolutional networks and applications in regulatory genomics and imaging, with practical recommendations. Predates transformers and language models. (review; CC BY 4.0)

  Deep learning for computational biology. Angermueller et al. (2016). Molecular Systems Biology. doi:10.15252/msb.20156651

  Abstract

  Technological advances in genomics and imaging have led to an explosion of molecular and cellular profiling data from large numbers of samples. This rapid increase in biological data dimension and acquisition rate is challenging conventional analysis strategies. Modern machine learning methods, such as deep learning, promise to leverage very large data sets for finding hidden structure within them, and for making accurate predictions. In this review, we discuss applications of this new breed of analysis approaches in regulatory genomics and cellular imaging. We provide background of what deep learning is, and the settings in which it can be successfully applied to derive biological insights. In addition to presenting specific applications and providing tips for practical use, we also highlight possible pitfalls and limitations to guide computational biologists when and how to make the most use of this new technology.

• Lee et al. (2022) — Short, practical guidance on when deep learning is appropriate, comparing with simple baselines, avoiding overfitting and leakage, and interpreting and sharing models. Complements the subsections on training and on interpretability and reporting. (tutorial; CC BY 4.0)

  Ten quick tips for deep learning in biology. Lee et al. (2022). PLOS Computational Biology. doi:10.1371/journal.pcbi.1009803

• Zhang et al. (2023) — Explains the transformer architecture and self-supervised pre-training, and surveys their use for protein, DNA and RNA sequences and other biological data. Relevant to the subsection on language models for proteins and DNA. (review; CC BY 4.0)

  Applications of transformer-based language models in bioinformatics: a survey. Zhang et al. (2023). Bioinformatics Advances. doi:10.1093/bioadv/vbad001

  Abstract

  The transformer-based language models, including vanilla transformer, BERT and GPT-3, have achieved revolutionary breakthroughs in the field of natural language processing (NLP). Since there are inherent similarities between various biological sequences and natural languages, the remarkable interpretability and adaptability of these models have prompted a new wave of their application in bioinformatics research. To provide a timely and comprehensive review, we introduce key developments of transformer-based language models by describing the detailed structure of transformers and summarize their contribution to a wide range of bioinformatics research from basic sequence analysis to drug discovery. While transformer-based applications in bioinformatics are diverse and multifaceted, we identify and discuss the common challenges, including heterogeneity of training data, computational expense and model interpretability, and opportunities in the context of bioinformatics research. We hope that the broader community of NLP researchers, bioinformaticians and biologists will be brought together to foster future research and development in transformer-based language models, and inspire novel bioinformatics applications that are unattainable by traditional methods. Supplementary information Supplementary data are available at Bioinformatics Advances online.

D.12.4 Generative models

• Lopez et al. (2020) — Explains the principles of deep generative models, in particular variational autoencoders and GANs, and how they are used for single-cell data, sequences and images. Predates the use of diffusion models in biology. (review; CC BY 4.0)

  Enhancing scientific discoveries in molecular biology with deep generative models. Lopez et al. (2020). Molecular Systems Biology. doi:10.15252/msb.20199198

  Abstract

  Generative models provide a well-established statistical framework for evaluating uncertainty and deriving conclusions from large data sets especially in the presence of noise, sparsity, and bias. Initially developed for computer vision and natural language processing, these models have been shown to effectively summarize the complexity that underlies many types of data and enable a range of applications including supervised learning tasks, such as assigning labels to images; unsupervised learning tasks, such as dimensionality reduction; and out-of-sample generation, such as de novo image synthesis. With this early success, the power of generative models is now being increasingly leveraged in molecular biology, with applications ranging from designing new molecules with properties of interest to identifying deleterious mutations in our genomes and to dissecting transcriptional variability between single cells. In this review, we provide a brief overview of the technical notions behind generative models and their implementation with deep learning techniques. We then describe several different ways in which these models can be utilized in practice, using several recent applications in molecular biology as examples.

• Strokach and Kim (2022) — A concise review of deep generative models of protein sequences and structures and how they are used, together with predictive models, to propose and select new protein designs. (review; CC BY 4.0)

  Deep generative modeling for protein design. Strokach & Kim (2022). Current Opinion in Structural Biology. doi:10.1016/j.sbi.2021.11.008

  Abstract

  Deep learning approaches have produced substantial breakthroughs in fields such as image classification and natural language processing and are making rapid inroads in the area of protein design. Many generative models of proteins have been developed that encompass all known protein sequences, model specific protein families, or extrapolate the dynamics of individual proteins. Those generative models can learn protein representations that are often more informative of protein structure and function than hand-engineered features. Furthermore, they can be used to quickly propose millions of novel proteins that resemble the native counterparts in terms of expression level, stability, or other attributes. The protein design process can further be guided by discriminative oracles to select candidates with the highest probability of having the desired properties. In this review, we discuss five classes of generative models that have been most successful at modeling proteins and provide a framework for model guided protein design.

• Mardikoraem et al. (2023) — Describes the architecture and mathematics of self-supervised and generative models for protein sequences and their use in protein engineering and fitness prediction. (review; CC BY 4.0)

  Generative models for protein sequence modeling: recent advances and future directions. Mardikoraem et al. (2023). Briefings in Bioinformatics. doi:10.1093/bib/bbad358

  Abstract

  The widespread adoption of high-throughput omics technologies has exponentially increased the amount of protein sequence data involved in many salient disease pathways and their respective therapeutics and diagnostics. Despite the availability of large-scale sequence data, the lack of experimental fitness annotations underpins the need for self-supervised and unsupervised machine learning (ML) methods. These techniques leverage the meaningful features encoded in abundant unlabeled sequences to accomplish complex protein engineering tasks. Proficiency in the rapidly evolving fields of protein engineering and generative AI is required to realize the full potential of ML models as a tool for protein fitness landscape navigation. Here, we support this work by (i) providing an overview of the architecture and mathematical details of the most successful ML models applicable to sequence data (e.g. variational autoencoders, autoregressive models, generative adversarial neural networks, and diffusion models), (ii) guiding how to effectively implement these models on protein sequence data to predict fitness or generate high-fitness sequences and (iii) highlighting several successful studies that implement these techniques in protein engineering (from paratope regions and subcellular localization prediction to high-fitness sequences and protein design rules generation). By providing a comprehensive survey of model details, novel architecture developments, comparisons of model applications, and current challenges, this study intends to provide structured guidance and robust framework for delivering a prospective outlook in the ML-driven protein engineering field.

D.12.5 Foundation models

• Li et al. (2024) — Reviews foundation models for sequences, structures and omics data, how they are pre-trained and adapted to downstream tasks, and their current limitations. (review; CC BY 4.0)

  Progress and opportunities of foundation models in bioinformatics. Li et al. (2024). Briefings in Bioinformatics. doi:10.1093/bib/bbae548

  Abstract

  Bioinformatics has undergone a paradigm shift in artificial intelligence (AI), particularly through foundation models (FMs), which address longstanding challenges in bioinformatics such as limited annotated data and data noise. These AI techniques have demonstrated remarkable efficacy across various downstream validation tasks, effectively representing diverse biological entities and heralding a new era in computational biology. The primary goal of this survey is to conduct a general investigation and summary of FMs in bioinformatics, tracing their evolutionary trajectory, current research landscape, and methodological frameworks. Our primary focus is on elucidating the application of FMs to specific biological problems, offering insights to guide the research community in choosing appropriate FMs for tasks like sequence analysis, structure prediction, and function annotation. Each section delves into the intricacies of the targeted challenges, contrasting the architectures and advancements of FMs with conventional methods and showcasing their utility across different biological domains. Further, this review scrutinizes the hurdles and constraints encountered by FMs in biology, including issues of data noise, model interpretability, and potential biases. This analysis provides a theoretical groundwork for understanding the circumstances under which certain FMs may exhibit suboptimal performance. Lastly, we outline prospective pathways and methodologies for the future development of FMs in biological research, facilitating ongoing innovation in the field. This comprehensive examination not only serves as an academic reference but also as a roadmap for forthcoming explorations and applications of FMs in biology.

• Guo et al. (2025) — A broad overview of language and foundation models for proteins, nucleic acids and single-cell data, with discussion of evaluation and open problems. (review; CC BY 4.0)

  Foundation models in bioinformatics. Guo et al. (2025). National Science Review. doi:10.1093/nsr/nwaf028

  Abstract

  With the adoption of foundation models (FMs), artificial intelligence (AI) has become increasingly significant in bioinformatics and has successfully addressed many historical challenges, such as pre-training frameworks, model evaluation and interpretability. FMs demonstrate notable proficiency in managing large-scale, unlabeled datasets, because experimental procedures are costly and labor intensive. In various downstream tasks, FMs have consistently achieved noteworthy results, demonstrating high levels of accuracy in representing biological entities. A new era in computational biology has been ushered in by the application of FMs, focusing on both general and specific biological issues. In this review, we introduce recent advancements in bioinformatics FMs employed in a variety of downstream tasks, including genomics, transcriptomics, proteomics, drug discovery and single-cell analysis. Our aim is to assist scientists in selecting appropriate FMs in bioinformatics, according to four model types: language FMs, vision FMs, graph FMs and multimodal FMs. In addition to understanding molecular landscapes, AI technology can establish the theoretical and practical foundation for continued innovation in molecular biology.

• Baek et al. (2025) — Explains how single-cell foundation models, typically transformers pre-trained on large collections of transcriptomes, are built and used, and discusses their current challenges. (review; CC BY 4.0)

  Single-cell foundation models: bringing artificial intelligence into cell biology. Baek et al. (2025). Experimental & Molecular Medicine. doi:10.1038/s12276-025-01547-5

  Abstract

  A foundation model, a large-scale deep learning model pretrained on vast datasets, has revolutionized data interpretation through self-supervised learning with capacity for various downstream tasks. Concurrently, single-cell genomics is in urgent need of unified frameworks capable of integrating and comprehensively analyzing the rapidly expanding data repositories. Inspired by advances in foundation models, researchers have extended these techniques to single-cell analysis, giving rise to single-cell foundation models (scFMs). Typically, these models use transformer architectures to incorporate diverse omics data and extract latent patterns at both cell and gene/feature levels for the analysis of cellular heterogeneity and complex regulatory networks. Despite their promise, scFMs face challenges including the nonsequential nature of omics data, inconsistency in data quality and the computational intensity required for training and fine-tuning. Furthermore, interpreting the biological relevance of latent embeddings and model representations remains nontrivial. Here we provide an overview of scFMs, highlighting their key concepts and applications across downstream tasks. We critically assess the current limitations and propose future directions aimed at enhancing the robustness, interpretability and scalability of scFMs. Ultimately, addressing these challenges will be crucial for establishing scFMs as pivotal tools in advancing single-cell genomics and unlocking deeper insights into cellular function and disease mechanisms.

D.12.6 AI in research practice

• Smith et al. (2024) — Practical rules for using LLMs responsibly in research, including verifying output, protecting data, disclosure and reproducibility. Matches the risks listed in the section. (tutorial; CC BY 4.0)

  Ten simple rules for using large language models in science, version 1.0. Smith et al. (2024). PLOS Computational Biology. doi:10.1371/journal.pcbi.1011767

  Abstract

  Box 1-An LLM suggests 10 simple rules governing its own usage in scienceHere, we ask ChatGPT to generate a list of 10 simple rules for using LLMs in science in order to illustrate its ability to provide detailed, human-like responses to short prompts.Q1: Can you provide me with a list of ten simple rules for using large language models in science?R1 (ChatGPT 3.5): Certainly!Here are ten simple rules for using large language models in science:1. Understand the limitations: Recognize that large language models like GPT-3 are powerful but not infallible.They may produce accurate-sounding information, but they can also generate false or biased content.Exercise caution and critical thinking.2. Validate and verify: Always cross-reference information generated by language models with reliable sources.Verify the accuracy of claims, check for

• Lubiana et al. (2023) — Advice on using chatbots for coding and other tasks in computational biology, with emphasis on checking generated code and text. (tutorial; CC BY 4.0)

  Ten quick tips for harnessing the power of ChatGPT in computational biology. Lubiana et al. (2023). PLOS Computational Biology. doi:10.1371/journal.pcbi.1011319

  Abstract

  The rise of advanced chatbots, such as ChatGPT, has stirred excitement and curiosity in the scientific community.Powered by large language models (LLMs) based on generative pretrained transformers (GPTs)-specifically GPT-3.5 and GPT-4-ChatGPT is considered a general-purpose technology with the potential to impact the job market and research endeavors in numerous fields [1].Although similar models have been fine-tuned for biology-specific projects, including text-based analysis and biological sequence decoding [2,3], ChatGPT provides a natural interface for bioinformaticians to begin using LLMs in their activities.This tool is already accelerating various activities undertaken by computational biologists, ranging from data cleaning to interpreting results and publishing.However, with great power comes great responsibility.As scientists, we must harness the full potential of ChatGPT while adhering to ethical guidelines and avoiding pitfalls associated with the technology.Here, we provide 10 insightful tips designed to help computational biologists optimize their workflows with ChatGPT, ranging from basic prompts to more advanced techniques.Although our primary focus is on the current ChatGPT/GPT-4 model, we believe that these tips will remain relevant for future iterations of the technology, as well as other LLMs and chatbots (such as Meta's LLaMa and Google's Bard) [4,5].We invite you to explore our 10 tips (summarized in Fig 1) aimed at effectively utilizing ChatGPT to advance computational biology research while maintaining a strong commitment to research integrity. Tip 1: Embrace the technology and be ready for noveltyChatGPT, a powerful tool for coding and academic writing tasks, is rapidly gaining traction in the scientific community.While exercising critical judgment and not blindly accepting everything it produces is important, incorporating ChatGPT into your workflow can undoubtedly

• Helmy et al. (2025) — Recent guidance on using generative AI tools in scientific work, covering verification, transparency and ethical issues. (tutorial; CC BY 4.0)

  Ten simple rules for optimal and careful use of generative AI in science. Helmy et al. (2025). PLOS Computational Biology. doi:10.1371/journal.pcbi.1013588

  Abstract

  Modern AI technologies leverage natural language processing (NLP), a subfield of AI dedicated to understanding, interpreting, and generating human language for developing large language models (LLMs), which have significantly advanced the capabilities of AI systems. These models can perform complex language tasks such as text generation, summarization, translation, and sentiment analysis, with unprecedented accuracy. The two main kinds of pre-training LLMs are the BERT-like models (e.g., BioBERT, proteinBERT, and PubMedBERT used primarily for language understanding; and the GPT-like models (e.g., BioGPT and ChatGPT-4o) used primarily for language generation

D.12.7 Integrating data

• Subramanian et al. (2020) — An overview of strategies and tools for integrating multi-omics data, organised by the question being asked (disease subtyping, biomarker prediction, insight into disease biology). (review; CC BY-NC 4.0)

  Multi-omics Data Integration, Interpretation, and Its Application. Subramanian et al. (2020). Bioinformatics and Biology Insights. doi:10.1177/1177932219899051

  Abstract

  To study complex biological processes holistically, it is imperative to take an integrative approach that combines multi-omics data to highlight the interrelationships of the involved biomolecules and their functions. With the advent of high-throughput techniques and availability of multi-omics data generated from a large set of samples, several promising tools and methods have been developed for data integration and interpretation. In this review, we collected the tools and methods that adopt integrative approach to analyze multiple omics data and summarized their ability to address applications such as disease subtyping, biomarker prediction, and deriving insights into the data. We provide the methodology, use-cases, and limitations of these tools; brief account of multi-omics data repositories and visualization portals; and challenges associated with multi-omics data integration.

• Argelaguet et al. (2018) — The MOFA paper; a clear explanation of how latent factors shared across data types are learned and interpreted. (primary study; CC BY 4.0)

  Multi-Omics Factor Analysis—a framework for unsupervised integration of multi-omics data sets. Argelaguet et al. (2018). Molecular Systems Biology. doi:10.15252/msb.20178124

  Abstract

  Multi-omics studies promise the improved characterization of biological processes across molecular layers. However, methods for the unsupervised integration of the resulting heterogeneous data sets are lacking. We present Multi-Omics Factor Analysis (MOFA), a computational method for discovering the principal sources of variation in multi-omics data sets. MOFA infers a set of (hidden) factors that capture biological and technical sources of variability. It disentangles axes of heterogeneity that are shared across multiple modalities and those specific to individual data modalities. The learnt factors enable a variety of downstream analyses, including identification of sample subgroups, data imputation and the detection of outlier samples. We applied MOFA to a cohort of 200 patient samples of chronic lymphocytic leukaemia, profiled for somatic mutations, RNA expression, DNA methylation and ex vivo drug responses. MOFA identified major dimensions of disease heterogeneity, including immunoglobulin heavy-chain variable region status, trisomy of chromosome 12 and previously underappreciated drivers, such as response to oxidative stress. In a second application, we used MOFA to analyse single-cell multi-omics data, identifying coordinated transcriptional and epigenetic changes along cell differentiation.

D.13 Image analysis and models of cells

Further explanations for Chapter 13.

D.13.1 Images as data

• Bankhead (2022) — An open textbook on what digital images are, how they are processed and measured, and how to avoid common mistakes; the best starting point for the classical part of image analysis. (book chapter; CC BY 4.0)

  Introduction to Bioimage Analysis. Bankhead (2022). Online book. https://bioimagebook.github.io/

• Moore et al. (2023) — Explains why large image data need new file formats and metadata standards, and how OME-Zarr is used to share them. (review; CC BY 4.0)

  OME-Zarr: a cloud-optimized bioimaging file format with international community support. Moore et al. (2023). Histochemistry and Cell Biology. doi:10.1007/s00418-023-02209-1

  Abstract

  A growing community is constructing a next-generation file format (NGFF) for bioimaging to overcome problems of scalability and heterogeneity. Organized by the Open Microscopy Environment (OME), individuals and institutes across diverse modalities facing these problems have designed a format specification process (OME-NGFF) to address these needs. This paper brings together a wide range of those community members to describe the cloud-optimized format itself-OME-Zarr-along with tools and data resources available today to increase FAIR access and remove barriers in the scientific process. The current momentum offers an opportunity to unify a key component of the bioimaging domain-the file format that underlies so many personal, institutional, and global data management and analysis tasks.

D.13.2 Deep learning for images

• Hallou et al. (2021) — A clear introduction to classification, segmentation, tracking and restoration with deep learning, written for biologists, with practical advice on training data and validation. (review; CC BY 4.0)

  Deep learning for bioimage analysis in developmental biology. Hallou et al. (2021). Development. doi:10.1242/dev.199616

  Abstract

  Deep learning has transformed the way large and complex image datasets can be processed, reshaping what is possible in bioimage analysis. As the complexity and size of bioimage data continues to grow, this new analysis paradigm is becoming increasingly ubiquitous. In this Review, we begin by introducing the concepts needed for beginners to understand deep learning. We then review how deep learning has impacted bioimage analysis and explore the open-source resources available to integrate it into a research project. Finally, we discuss the future of deep learning applied to cell and developmental biology. We analyze how state-of-the-art methodologies have the potential to transform our understanding of biological systems through new image-based analysis and modelling that integrate multimodal inputs in space and time.

• Chamier et al. (2021) — Describes notebooks for training common networks (U-Net, StarDist, CARE and others) without installing software; useful for trying the methods yourself. (tutorial; CC BY 4.0)

  Democratising deep learning for microscopy with ZeroCostDL4Mic. von Chamier et al. (2021). Nature Communications. doi:10.1038/s41467-021-22518-0

  Abstract

  Deep Learning (DL) methods are powerful analytical tools for microscopy and can outperform conventional image processing pipelines. Despite the enthusiasm and innovations fuelled by DL technology, the need to access powerful and compatible resources to train DL networks leads to an accessibility barrier that novice users often find difficult to overcome. Here, we present ZeroCostDL4Mic, an entry-level platform simplifying DL access by leveraging the free, cloud-based computational resources of Google Colab. ZeroCostDL4Mic allows researchers with no coding expertise to train and apply key DL networks to perform tasks including segmentation (using U-Net and StarDist), object detection (using YOLOv2), denoising (using CARE and Noise2Void), super-resolution microscopy (using Deep-STORM), and image-to-image translation (using Label-free prediction - fnet, pix2pix and CycleGAN). Importantly, we provide suitable quantitative tools for each network to evaluate model performance, allowing model optimisation. We demonstrate the application of the platform to study multiple biological processes.

D.13.3 Image-based profiling

• Caicedo et al. (2017) — The standard description of the profiling workflow, from illumination correction and feature extraction to normalisation, batch correction and similarity measures. (review; CC BY 4.0)

  Data-analysis strategies for image-based cell profiling. Caicedo et al. (2017). Nature Methods. doi:10.1038/nmeth.4397

  Abstract

  Image-based cell profiling is a high-throughput strategy for the quantification of phenotypic differences among a variety of cell populations. It paves the way to studying biological systems on a large scale by using chemical and genetic perturbations. The general workflow for this technology involves image acquisition with high-throughput microscopy systems and subsequent image processing and analysis. Here, we introduce the steps required to create high-quality image-based (i.e., morphological) profiles from a collection of microscopy images. We recommend techniques that have proven useful in each stage of the data analysis process, on the basis of the experience of 20 laboratories worldwide that are refining their image-based cell-profiling methodologies in pursuit of biological discovery. The recommended techniques cover alternatives that may suit various biological goals, experimental designs, and laboratories' preferences.

• Stirling et al. (2021) — Describes the most widely used software for extracting features from Cell Painting and other screens. (tutorial; CC BY 4.0)

  CellProfiler 4: improvements in speed, utility and usability. Stirling et al. (2021). BMC Bioinformatics. doi:10.1186/s12859-021-04344-9

  Abstract

  Background Imaging data contains a substantial amount of information which can be difficult to evaluate by eye. With the expansion of high throughput microscopy methodologies producing increasingly large datasets, automated and objective analysis of the resulting images is essential to effectively extract biological information from this data. CellProfiler is a free, open source image analysis program which enables researchers to generate modular pipelines with which to process microscopy images into interpretable measurements. Results Herein we describe CellProfiler 4, a new version of this software with expanded functionality. Based on user feedback, we have made several user interface refinements to improve the usability of the software. We introduced new modules to expand the capabilities of the software. We also evaluated performance and made targeted optimizations to reduce the time and cost associated with running common large-scale analysis pipelines. Conclusions CellProfiler 4 provides significantly improved performance in complex workflows compared to previous versions. This release will ensure that researchers will have continued access to CellProfiler's powerful computational tools in the coming years.

D.13.4 Protein localisation from images

• Xiao et al. (2024) — Reviews both sequence-based and image-based prediction of localisation, and so connects this section with the sequence-based methods of chapter 10. (review; CC BY 4.0)

  A Review for Artificial Intelligence Based Protein Subcellular Localization. Xiao et al. (2024). Biomolecules. doi:10.3390/biom14040409

  Abstract

  Proteins need to be located in appropriate spatiotemporal contexts to carry out their diverse biological functions. Mislocalized proteins may lead to a broad range of diseases, such as cancer and Alzheimer's disease. Knowing where a target protein resides within a cell will give insights into tailored drug design for a disease. As the gold validation standard, the conventional wet lab uses fluorescent microscopy imaging, immunoelectron microscopy, and fluorescent biomarker tags for protein subcellular location identification. However, the booming era of proteomics and high-throughput sequencing generates tons of newly discovered proteins, making protein subcellular localization by wet-lab experiments a mission impossible. To tackle this concern, in the past decades, artificial intelligence (AI) and machine learning (ML), especially deep learning methods, have made significant progress in this research area. In this article, we review the latest advances in AI-based method development in three typical types of approaches, including sequence-based, knowledge-based, and image-based methods. We also elaborately discuss existing challenges and future directions in AI-based method development in this research field.

• Kobayashi et al. (2022) — A readable example of how self-supervised learning on images groups proteins by localisation without labels (cytoself). (primary study; CC BY 4.0)

  Self-supervised deep learning encodes high-resolution features of protein subcellular localization. Kobayashi et al. (2022). Nature Methods. doi:10.1038/s41592-022-01541-z

  Abstract

  Explaining the diversity and complexity of protein localization is essential to fully understand cellular architecture. Here we present cytoself, a deep-learning approach for fully self-supervised protein localization profiling and clustering. Cytoself leverages a self-supervised training scheme that does not require preexisting knowledge, categories or annotations. Training cytoself on images of 1,311 endogenously labeled proteins from the OpenCell database reveals a highly resolved protein localization atlas that recapitulates major scales of cellular organization, from coarse classes, such as nuclear and cytoplasmic, to the subtle localization signatures of individual protein complexes. We quantitatively validate cytoself's ability to cluster proteins into organelles and protein complexes, showing that cytoself outperforms previous self-supervised approaches. Moreover, to better understand the inner workings of our model, we dissect the emergent features from which our clustering is derived, interpret them in the context of the fluorescence images, and analyze the performance contributions of each component of our approach.

D.13.5 Tissues: multiplexed imaging and computational pathology

• Park et al. (2022) — An overview of imaging- and sequencing-based spatial methods for RNA, proteins and other molecules, and how they are combined. (review; CC BY 4.0)

  Spatial omics technologies at multimodal and single cell/subcellular level. Park et al. (2022). Genome Biology. doi:10.1186/s13059-022-02824-6

  Abstract

  Spatial omics technologies enable a deeper understanding of cellular organizations and interactions within a tissue of interest. These assays can identify specific compartments or regions in a tissue with differential transcript or protein abundance, delineate their interactions, and complement other methods in defining cellular phenotypes. A variety of spatial methodologies are being developed and commercialized; however, these techniques differ in spatial resolution, multiplexing capability, scale/throughput, and coverage. Here, we review the current and prospective landscape of single cell to subcellular resolution spatial omics technologies and analysis tools to provide a comprehensive picture for both research and clinical applications.

• Howard et al. (2021) — Shows how hidden differences between hospitals inflate the apparent accuracy of pathology models; a good case study of data leakage. (primary study; CC BY 4.0)

  The impact of site-specific digital histology signatures on deep learning model accuracy and bias. Howard et al. (2021). Nature Communications. doi:10.1038/s41467-021-24698-1

  Abstract

  The Cancer Genome Atlas (TCGA) is one of the largest biorepositories of digital histology. Deep learning (DL) models have been trained on TCGA to predict numerous features directly from histology, including survival, gene expression patterns, and driver mutations. However, we demonstrate that these features vary substantially across tissue submitting sites in TCGA for over 3,000 patients with six cancer subtypes. Additionally, we show that histologic image differences between submitting sites can easily be identified with DL. Site detection remains possible despite commonly used color normalization and augmentation methods, and we quantify the image characteristics constituting this site-specific digital histology signature. We demonstrate that these site-specific signatures lead to biased accuracy for prediction of features including survival, genomic mutations, and tumor stage. Furthermore, ethnicity can also be inferred from site-specific signatures, which must be accounted for to ensure equitable application of DL. These site-specific signatures can lead to overoptimistic estimates of model performance, and we propose a quadratic programming method that abrogates this bias by ensuring models are not trained and validated on samples from the same site.

D.13.6 Cell atlases and virtual cells

• Regev et al. (2017) — The white paper that defined the aims of the Human Cell Atlas, including what is meant by a cell type and why images and locations are part of it. (perspective; CC BY 4.0)

  The Human Cell Atlas. Regev et al. (2017). eLife. doi:10.7554/elife.27041

  Abstract

  The recent advent of methods for high-throughput single-cell molecular profiling has catalyzed a growing sense in the scientific community that the time is ripe to complete the 150-year-old effort to identify all cell types in the human body. The Human Cell Atlas Project is an international collaborative effort that aims to define all human cell types in terms of distinctive molecular profiles (such as gene expression profiles) and to connect this information with classical cellular descriptions (such as location and morphology). An open comprehensive reference map of the molecular state of cells in healthy human tissues would propel the systematic study of physiological states, developmental trajectories, regulatory circuitry and interactions of cells, and also provide a framework for understanding cellular dysregulation in human disease. Here we describe the idea, its potential utility, early proofs-of-concept, and some design considerations for the Human Cell Atlas, including a commitment to open data, code, and community.

• Kedzierska et al. (2025) — A careful evaluation of single-cell foundation models against simple baselines; useful for discussing what virtual cell models can and cannot yet do. (primary study; CC BY 4.0)

  Zero-shot evaluation reveals limitations of single-cell foundation models. Kedzierska et al. (2025). Genome Biology. doi:10.1186/s13059-025-03574-x

  Abstract

  Foundation models such as scGPT and Geneformer have not been rigorously evaluated in a setting where they are used without any further training (i.e., zero-shot). Understanding the performance of models in zero-shot settings is critical to applications that exclude the ability to fine-tune, such as discovery settings where labels are unknown. Our evaluation of the zero-shot performance of Geneformer and scGPT suggests that, in some cases, these models may face reliability challenges and could be outperformed by simpler methods. Our findings underscore the importance of zero-shot evaluations in development and deployment of foundation models in single-cell research.

D.13.7 Whole-cell models and cryo-electron tomography

• Young and Villa (2023) — An open review of cryo-electron tomography, from sample thinning by focused ion beams to subtomogram averaging and visual proteomics. (review; CC BY 4.0)

  Bringing Structure to Cell Biology with Cryo-Electron Tomography. Young & Villa (2023). Annual Review of Biophysics. doi:10.1146/annurev-biophys-111622-091327

  Abstract

  Recent advances in cryo-electron microscopy have marked only the beginning of the potential of this technique. To bring structure into cell biology, the modality of cryo-electron tomography has fast developed into a bona fide in situ structural biology technique where structures are determined in their native environment, the cell. Nearly every step of the cryo-focused ion beam-assisted electron tomography (cryo-FIB-ET) workflow has been improved upon in the past decade, since the first windows were carved into cells, unveiling macromolecular networks in near-native conditions. By bridging structural and cell biology, cryo-FIB-ET is advancing our understanding of structure-function relationships in their native environment and becoming a tool for discovering new biology.

• Stevens et al. (2023) — Explains how a whole minimal cell could be built and simulated with coarse-grained molecular dynamics, and what data this requires. (perspective; CC BY 4.0)

  Molecular dynamics simulation of an entire cell. Stevens et al. (2023). Frontiers in Chemistry. doi:10.3389/fchem.2023.1106495

  Abstract

  The ultimate microscope, directed at a cell, would reveal the dynamics of all the cell's components with atomic resolution. In contrast to their real-world counterparts, computational microscopes are currently on the brink of meeting this challenge. In this perspective, we show how an integrative approach can be employed to model an entire cell, the minimal cell, JCVI-syn3A, at full complexity. This step opens the way to interrogate the cell's spatio-temporal evolution with molecular dynamics simulations, an approach that can be extended to other cell types in the near future.

References

Abriata, Luciano A., and Matteo Dal Peraro. 2025. “Practical Outcomes from CASP16 for Users in Need of Biomolecular Structure Prediction.” Proteins: Structure, Function, and Bioinformatics 94 (1): 435–46. https://doi.org/10.1002/prot.70078.

Adames, Neil R., Jenna E. Gallegos, and Jean Peccoud. 2018. “Yeast Genetic Interaction Screens in the Age of CRISPR/Cas.” Current Genetics 65 (2): 307–27. https://doi.org/10.1007/s00294-018-0887-8.

Ahlquist, K D, Mayra M Bañuelos, Alyssa Funk, et al. 2021. “Our Tangled Family Tree: New Genomic Methods Offer Insight into the Legacy of Archaic Admixture.” Genome Biology and Evolution 13 (7). https://doi.org/10.1093/gbe/evab115.

Alser, Mohammed, Jeremy Rotman, Dhrithi Deshpande, et al. 2021. “Technology Dictates Algorithms: Recent Developments in Read Alignment.” Genome Biology 22 (1). https://doi.org/10.1186/s13059-021-02443-7.

Amarasinghe, Shanika L., Shian Su, Xueyi Dong, Luke Zappia, Matthew E. Ritchie, and Quentin Gouil. 2020. “Opportunities and Challenges in Long-Read Sequencing Data Analysis.” Genome Biology 21 (1). https://doi.org/10.1186/s13059-020-1935-5.

Angelis, Jesse, Eva Ayla Schröder, Zixuan Xiao, Wassim Gabriel, and Mathias Wilhelm. 2025. “Peptide Property Prediction for Mass Spectrometry Using AI: An Introduction to State of the Art Models.” PROTEOMICS 25 (9-10). https://doi.org/10.1002/pmic.202400398.

Angermueller, Christof, Tanel Pärnamaa, Leopold Parts, and Oliver Stegle. 2016. “Deep Learning for Computational Biology.” Molecular Systems Biology 12 (7). https://doi.org/10.15252/msb.20156651.

Argelaguet, Ricard, Britta Velten, Damien Arnol, et al. 2018. “Multi-Omics Factor Analysis—a Framework for Unsupervised Integration of Multi-Omics Data Sets.” Molecular Systems Biology 14 (6). https://doi.org/10.15252/msb.20178124.

Aris-Brosou, Stéphane, and Nicolas Rodrigue. 2019. “A Not-so-Long Introduction to Computational Molecular Evolution.” In Evolutionary Genomics. Springer New York. https://doi.org/10.1007/978-1-4939-9074-0_3.

Armstrong, Joel, Ian T. Fiddes, Mark Diekhans, and Benedict Paten. 2019. “Whole-Genome Alignment and Comparative Annotation.” Annual Review of Animal Biosciences 7 (1): 41–64. https://doi.org/10.1146/annurev-animal-020518-115005.

Baek, Seungbyn, Kyungwoo Song, and Insuk Lee. 2025. “Single-Cell Foundation Models: Bringing Artificial Intelligence into Cell Biology.” Experimental & Molecular Medicine 57 (10): 2169–81. https://doi.org/10.1038/s12276-025-01547-5.

Bankhead, Peter. 2022. Introduction to Bioimage Analysis. Online book. https://bioimagebook.github.io/.

Beerenwinkel, Niko, Roland F. Schwarz, Moritz Gerstung, and Florian Markowetz. 2014. “Cancer Evolution: Mathematical Models and Computational Inference.” Systematic Biology 64 (1): e1–25. https://doi.org/10.1093/sysbio/syu081.

Ben-Hur, Asa, Cheng Soon Ong, Sören Sonnenburg, Bernhard Schölkopf, and Gunnar Rätsch. 2008. “Support Vector Machines and Kernels for Computational Biology.” PLoS Computational Biology 4 (10): e1000173. https://doi.org/10.1371/journal.pcbi.1000173.

Bertoline, Letícia M. F., Angélica N. Lima, Jose E. Krieger, and Samantha K. Teixeira. 2023. “Before and After AlphaFold2: An Overview of Protein Structure Prediction.” Frontiers in Bioinformatics 3. https://doi.org/10.3389/fbinf.2023.1120370.

Birchler, James A, and Hua Yang. 2022. “The Multiple Fates of Gene Duplications: Deletion, Hypofunctionalization, Subfunctionalization, Neofunctionalization, Dosage Balance Constraints, and Neutral Variation.” The Plant Cell 34 (7): 2466–74. https://doi.org/10.1093/plcell/koac076.

Birney, Ewan. 2021. “The International Human Genome Project.” Human Molecular Genetics 30 (R2): R161–63. https://doi.org/10.1093/hmg/ddab198.

Bjornson, Saelin, Heroen Verbruggen, Nathan S. Upham, and Jacob L. Steenwyk. 2024. “Reticulate Evolution: Detection and Utility in the Phylogenomics Era.” Molecular Phylogenetics and Evolution 201: 108197. https://doi.org/10.1016/j.ympev.2024.108197.

Bobay, Louis-Marie, and Howard Ochman. 2017. “The Evolution of Bacterial Genome Architecture.” Frontiers in Genetics 8. https://doi.org/10.3389/fgene.2017.00072.

Bock, Christoph, Paul Datlinger, Florence Chardon, et al. 2022. “High-Content CRISPR Screening.” Nature Reviews Methods Primers 2 (1). https://doi.org/10.1038/s43586-021-00093-4.

Bowman, Gregory R. 2024. “AlphaFold and Protein Folding: Not Dead yet! The Frontier Is Conformational Ensembles.” Annual Review of Biomedical Data Science 7 (1): 51–57. https://doi.org/10.1146/annurev-biodatasci-102423-011435.

Braun, Efrem, Justin Gilmer, Heather B. Mayes, et al. 2018. “Best Practices for Foundations in Molecular Simulations [Article V1.0].” Living Journal of Computational Molecular Science 1 (1): 5957. https://doi.org/10.33011/livecoms.1.1.5957.

Bravo, Gustavo A., Alexandre Antonelli, Christine D. Bacon, et al. 2019. “Embracing Heterogeneity: Coalescing the Tree of Life and the Future of Phylogenomics.” PeerJ 7: e6399. https://doi.org/10.7717/peerj.6399.

Burley, Stephen K., Helen M. Berman, Jose M. Duarte, et al. 2022. “Protein Data Bank: A Comprehensive Review of 3D Structure Holdings and Worldwide Utilization by Researchers, Educators, and Students.” Biomolecules 12 (10): 1425. https://doi.org/10.3390/biom12101425.

Burley, Stephen K., Dennis W. Piehl, Brinda Vallat, and Christine Zardecki. 2024. “RCSB Protein Data Bank: Supporting Research and Education Worldwide Through Explorations of Experimentally Determined and Computationally Predicted Atomic Level 3D Biostructures.” IUCrJ 11 (3): 279–86. https://doi.org/10.1107/s2052252524002604.

Caicedo, Juan C, Sam Cooper, Florian Heigwer, et al. 2017. “Data-Analysis Strategies for Image-Based Cell Profiling.” Nature Methods 14 (9): 849–63. https://doi.org/10.1038/nmeth.4397.

Casillas, Sònia, and Antonio Barbadilla. 2017. “Molecular Population Genetics.” Genetics 205 (3): 1003–35. https://doi.org/10.1534/genetics.116.196493.

Chamier, Lucas von, Romain F. Laine, Johanna Jukkala, et al. 2021. “Democratising Deep Learning for Microscopy with ZeroCostDL4Mic.” Nature Communications 12 (1). https://doi.org/10.1038/s41467-021-22518-0.

Chandra, Abel, Laura Tünnermann, Tommy Löfstedt, and Regina Gratz. 2023. “Transformer-Based Deep Learning for Predicting Protein Properties in the Life Sciences.” eLife 12. https://doi.org/10.7554/elife.82819.

Chicco, Davide. 2017. “Ten Quick Tips for Machine Learning in Computational Biology.” BioData Mining 10 (1). https://doi.org/10.1186/s13040-017-0155-3.

Chicco, Davide, and Giuseppe Agapito. 2022. “Nine Quick Tips for Pathway Enrichment Analysis.” PLOS Computational Biology 18 (8): e1010348. https://doi.org/10.1371/journal.pcbi.1010348.

Clark, Douglas, Choi (OpenStax). 2018a. Chapter 14: DNA Structure and Function (Biology 2e). OpenStax Biology 2e. https://openstax.org/books/biology-2e/pages/14-introduction.

Clark, Douglas, Choi (OpenStax). 2018b. Chapter 15: Genes and Proteins (Biology 2e). OpenStax Biology 2e. https://openstax.org/books/biology-2e/pages/15-introduction.

Cornish-Bowden, Athel. 2015. “One Hundred Years of Michaelis–Menten Kinetics.” Perspectives in Science 4: 3–9. https://doi.org/10.1016/j.pisc.2014.12.002.

Corpas, Manuel, Nadezda V. Kovalevskaya, Amanda McMurray, and Fiona G. G. Nielsen. 2018. “A FAIR Guide for Data Providers to Maximise Sharing of Human Genomic Data.” PLOS Computational Biology 14 (3): e1005873. https://doi.org/10.1371/journal.pcbi.1005873.

Cozzetto, Domenico, and David T. Jones. 2016. “Computational Methods for Annotation Transfers from Sequence.” In The Gene Ontology Handbook. Springer New York. https://doi.org/10.1007/978-1-4939-3743-1_5.

Dam, Sipko van, Urmo Võsa, Adriaan van der Graaf, Lude Franke, and João Pedro de Magalhães. 2017. “Gene Co-Expression Analysis for Functional Classification and Gene–Disease Predictions.” Briefings in Bioinformatics, bbw139. https://doi.org/10.1093/bib/bbw139.

Danielewski, Mikołaj, Marlena Szalata, Jan Krzysztof Nowak, Jarosław Walkowiak, Ryszard Słomski, and Karolina Wielgus. 2025. “History of Biological Databases, Their Importance, and Existence in Modern Scientific and Policy Context.” Genes 16 (1): 100. https://doi.org/10.3390/genes16010100.

Das, Sayoni, Natalie L Dawson, and Christine A Orengo. 2015. “Diversity in Protein Domain Superfamilies.” Current Opinion in Genetics & Development 35: 40–49. https://doi.org/10.1016/j.gde.2015.09.005.

David, Alessia, Suhail Islam, Evgeny Tankhilevich, and Michael J. E. Sternberg. 2022. “The AlphaFold Database of Protein Structures: A Biologist’s Guide.” Journal of Molecular Biology 434 (2): 167336. https://doi.org/10.1016/j.jmb.2021.167336.

De Las Rivas, Javier, and Celia Fontanillo. 2010. “Protein–Protein Interactions Essentials: Key Concepts to Building and Analyzing Interactome Networks.” PLoS Computational Biology 6 (6): e1000807. https://doi.org/10.1371/journal.pcbi.1000807.

Deutsch, Eric W., Juan Antonio Vizcaíno, Andrew R. Jones, et al. 2023. “Proteomics Standards Initiative at Twenty Years: Current Activities and Future Work.” Journal of Proteome Research 22 (2): 287–301. https://doi.org/10.1021/acs.jproteome.2c00637.

Dominguez Del Angel, Victoria, Erik Hjerde, Lieven Sterck, et al. 2018. “Ten Steps to Get Started in Genome Assembly and Annotation.” F1000Research 7: 148. https://doi.org/10.12688/f1000research.13598.1.

Dupree, Emmalyn J., Madhuri Jayathirtha, Hannah Yorkey, Marius Mihasan, Brindusa Alina Petre, and Costel C. Darie. 2020. “A Critical Review of Bottom-up Proteomics: The Good, the Bad, and the Future of This Field.” Proteomes 8 (3): 14. https://doi.org/10.3390/proteomes8030014.

Edfors, Fredrik, Andreas Hober, Klas Linderbäck, et al. 2018. “Enhanced Validation of Antibodies for Research Applications.” Nature Communications 9 (1). https://doi.org/10.1038/s41467-018-06642-y.

Ejigu, Girum Fitihamlak, and Jaehee Jung. 2020. “Review on the Computational Genome Annotation of Sequences Obtained by Next-Generation Sequencing.” Biology 9 (9): 295. https://doi.org/10.3390/biology9090295.

Ekblom, Robert, and Jochen B. W. Wolf. 2014. “A Field Guide to Whole-Genome Sequencing, Assembly and Annotation.” Evolutionary Applications 7 (9): 1026–42. https://doi.org/10.1111/eva.12178.

Eldjarn, Grimur Hjorleifsson, Egil Ferkingstad, Sigrun H. Lund, et al. 2023. “Large-Scale Plasma Proteomics Comparisons Through Genetics and Disease Associations.” Nature 622 (7982): 348–58. https://doi.org/10.1038/s41586-023-06563-x.

Elofsson, Arne. 2023. “Progress at Protein Structure Prediction, as Seen in CASP15.” Current Opinion in Structural Biology 80: 102594. https://doi.org/10.1016/j.sbi.2023.102594.

Fleming, Jennifer, Paulyna Magana, Sreenath Nair, et al. 2025. “AlphaFold Protein Structure Database and 3D-Beacons: New Data and Capabilities.” Journal of Molecular Biology 437 (15): 168967. https://doi.org/10.1016/j.jmb.2025.168967.

Forslund, Sofia K., Mateusz Kaduk, and Erik L. L. Sonnhammer. 2019. “Evolution of Protein Domain Architectures.” In Evolutionary Genomics. Springer New York. https://doi.org/10.1007/978-1-4939-9074-0_15.

Friedberg, Iddo, and Predrag Radivojac. 2016. “Community-Wide Evaluation of Computational Function Prediction.” In The Gene Ontology Handbook. Springer New York. https://doi.org/10.1007/978-1-4939-3743-1_10.

Garrido-Ramos, Manuel. 2017. “Satellite DNA: An Evolving Topic.” Genes 8 (9): 230. https://doi.org/10.3390/genes8090230.

Gaudet, Pascale, Nives Škunca, James C. Hu, and Christophe Dessimoz. 2016. “Primer on the Gene Ontology.” In The Gene Ontology Handbook. Springer New York. https://doi.org/10.1007/978-1-4939-3743-1_3.

Gerasimavicius, Lukas, Sarah A. Teichmann, and Joseph A. Marsh. 2025. “Leveraging Protein Structural Information to Improve Variant Effect Prediction.” Current Opinion in Structural Biology 92: 103023. https://doi.org/10.1016/j.sbi.2025.103023.

Gianni, Stefano, María Inés Freiberger, Per Jemth, Diego U. Ferreiro, Peter G. Wolynes, and Monika Fuxreiter. 2021. “Fuzziness and Frustration in the Energy Landscape of Protein Folding, Function, and Assembly.” Accounts of Chemical Research 54 (5): 1251–59. https://doi.org/10.1021/acs.accounts.0c00813.

Gibson, Greg, Joseph E Powell, and Urko M Marigorta. 2015. “Expression Quantitative Trait Locus Analysis for Translational Medicine.” Genome Medicine 7 (1). https://doi.org/10.1186/s13073-015-0186-7.

Gillani, Maryam, and Gianluca Pollastri. 2024. “Protein Subcellular Localization Prediction Tools <Sup/>.” Computational and Structural Biotechnology Journal 23: 1796–807. https://doi.org/10.1016/j.csbj.2024.04.032.

Glover, Natasha, Christophe Dessimoz, Ingo Ebersberger, et al. 2019. “Advances and Applications in the Quest for Orthologs.” Molecular Biology and Evolution 36 (10): 2157–64. https://doi.org/10.1093/molbev/msz150.

Gregory, T. Ryan. 2008. “Understanding Evolutionary Trees.” Evolution: Education and Outreach 1 (2): 121–37. https://doi.org/10.1007/s12052-008-0035-x.

Gu, Changdai, Gi Bae Kim, Won Jun Kim, Hyun Uk Kim, and Sang Yup Lee. 2019. “Current Status and Applications of Genome-Scale Metabolic Models.” Genome Biology 20 (1). https://doi.org/10.1186/s13059-019-1730-3.

Guo, Fei, Renchu Guan, Yaohang Li, et al. 2025. “Foundation Models in Bioinformatics.” National Science Review 12 (4). https://doi.org/10.1093/nsr/nwaf028.

Hallou, Adrien, Hannah G. Yevick, Bianca Dumitrascu, and Virginie Uhlmann. 2021. “Deep Learning for Bioimage Analysis in Developmental Biology.” Development 148 (18). https://doi.org/10.1242/dev.199616.

Haque, Ashraful, Jessica Engel, Sarah A. Teichmann, and Tapio Lönnberg. 2017. “A Practical Guide to Single-Cell RNA-Sequencing for Biomedical Research and Clinical Applications.” Genome Medicine 9 (1). https://doi.org/10.1186/s13073-017-0467-4.

Heather, James M., and Benjamin Chain. 2016. “The Sequence of Sequencers: The History of Sequencing DNA.” Genomics 107 (1): 1–8. https://doi.org/10.1016/j.ygeno.2015.11.003.

Helmy, Mohamed, Lingling Jin, Amr Alhossary, Tamer Mansour, Diogo Pellagrina, and Kumar Selvarajoo. 2025. “Ten Simple Rules for Optimal and Careful Use of Generative AI in Science.” PLOS Computational Biology 21 (10): e1013588. https://doi.org/10.1371/journal.pcbi.1013588.

Horne, Jim E., David J. Brockwell, and Sheena E. Radford. 2020. “Role of the Lipid Bilayer in Outer Membrane Protein Folding in Gram-Negative Bacteria.” Journal of Biological Chemistry 295 (30): 10340–67. https://doi.org/10.1074/jbc.rev120.011473.

Howard, Frederick M., James Dolezal, Sara Kochanny, et al. 2021. “The Impact of Site-Specific Digital Histology Signatures on Deep Learning Model Accuracy and Bias.” Nature Communications 12 (1). https://doi.org/10.1038/s41467-021-24698-1.

Hutchings, Charlotte, Charlotte S. Dawson, Thomas Krueger, Kathryn S. Lilley, and Lisa M. Breckels. 2023. “A Bioconductor Workflow for Processing, Evaluating, and Interpreting Expression Proteomics Data.” F1000Research 12: 1402. https://doi.org/10.12688/f1000research.139116.1.

Imai, Kenichiro, and Kenta Nakai. 2020. “Tools for the Recognition of Sorting Signals and the Prediction of Subcellular Localization of Proteins from Their Amino Acid Sequences.” Frontiers in Genetics 11. https://doi.org/10.3389/fgene.2020.607812.

Jacques, Florian, Paulina Bolivar, Kristian Pietras, and Emma U. Hammarlund. 2023. “Roadmap to the Study of Gene and Protein Phylogeny and Evolution—a Practical Guide.” PLOS ONE 18 (2): e0279597. https://doi.org/10.1371/journal.pone.0279597.

Jakubowski & Flatt. 2024. Amino Acids, Peptides, and Proteins (Chapter 3 of Fundamentals of Biochemistry). LibreTexts Biology (open textbook). https://bio.libretexts.org/Bookshelves/Biochemistry/Fundamentals_of_Biochemistry_(Jakubowski_and_Flatt)/01%3A_Unit_I-_Structure_and_Catalysis/03%3A_Amino_Acids_Peptides_and_Proteins.

Jänes, Jürgen, and Pedro Beltrao. 2024. “Deep Learning for Protein Structure Prediction and Design—Progress and Applications.” Molecular Systems Biology 20 (3): 162–69. https://doi.org/10.1038/s44320-024-00016-x.

Jarmoskaite, Inga, Ishraq AlSadhan, Pavanapuresan P Vaidyanathan, and Daniel Herschlag. 2020. “How to Measure and Evaluate Binding Affinities.” eLife 9. https://doi.org/10.7554/elife.57264.

Jiang, Yuming, Devasahayam Arokia Balaya Rex, Dina Schuster, et al. 2024. “Comprehensive Overview of Bottom-up Proteomics Using Mass Spectrometry.” ACS Measurement Science Au 4 (4): 338–417. https://doi.org/10.1021/acsmeasuresciau.3c00068.

Jones, Christopher T., Edward Susko, and Joseph P. Bielawski. 2019. “Looking for Darwin in Genomic Sequences: Validity and Success Depends on the Relationship Between Model and Data.” In Evolutionary Genomics. Springer New York. https://doi.org/10.1007/978-1-4939-9074-0_13.

Joy, Jeffrey B., Richard H. Liang, Rosemary M. McCloskey, T. Nguyen, and Art F. Y. Poon. 2016. “Ancestral Reconstruction.” PLOS Computational Biology 12 (7): e1004763. https://doi.org/10.1371/journal.pcbi.1004763.

Jung, Hyungtaek, Tomer Ventura, J. Sook Chung, et al. 2020. “Twelve Quick Steps for Genome Assembly and Annotation in the Classroom.” PLOS Computational Biology 16 (11): e1008325. https://doi.org/10.1371/journal.pcbi.1008325.

Kedzierska, Kasia Z., Lorin Crawford, Ava P. Amini, and Alex X. Lu. 2025. “Zero-Shot Evaluation Reveals Limitations of Single-Cell Foundation Models.” Genome Biology 26 (1). https://doi.org/10.1186/s13059-025-03574-x.

Kellis et al. 2016a. Inferring Orthologs/Paralogs, Gene Duplication and Loss. Computational Biology: Genomes, Networks, Evolution (MIT 6.047, LibreTexts). https://bio.libretexts.org/Bookshelves/Computational_Biology/Book%3A_Computational_Biology_-_Genomes_Networks_and_Evolution_(Kellis_et_al.)/27%3A_Phylogenomics_II/27.05%3A_Inferring_Orthologs_Paralogs_Gene_Duplication_and_Loss.

Kellis et al. 2016b. Natural Selection. Computational Biology: Genomes, Networks, Evolution (MIT 6.047, LibreTexts). https://bio.libretexts.org/Bookshelves/Computational_Biology/Book%3A_Computational_Biology_-_Genomes_Networks_and_Evolution_(Kellis_et_al.)/29%3A_Population_Genetic_Variation/29.04%3A_Natural_Selection.

Kellis et al. 2020a. Hidden Markov Models i (Chapter 7 of Computational Biology: Genomes, Networks, and Evolution). LibreTexts (MIT open textbook). https://bio.libretexts.org/Bookshelves/Computational_Biology/Book%3A_Computational_Biology_-_Genomes_Networks_and_Evolution_(Kellis_et_al.)/07%3A_Hidden_Markov_Models_I.

Kellis et al. 2020b. Rapid Sequence Alignment and Database Search (Chapter 3 of Computational Biology: Genomes, Networks, and Evolution). LibreTexts (MIT open textbook). https://bio.libretexts.org/Bookshelves/Computational_Biology/Book%3A_Computational_Biology_-_Genomes_Networks_and_Evolution_(Kellis_et_al.)/03%3A_Rapid_Sequence_Alignment_and_Database_Search.

Kellis et al. 2020c. RNA Folding (Chapter 10 of Computational Biology: Genomes, Networks, and Evolution). LibreTexts (MIT open textbook). https://bio.libretexts.org/Bookshelves/Computational_Biology/Book%3A_Computational_Biology_-_Genomes_Networks_and_Evolution_(Kellis_et_al.)/10%3A_RNA_Folding.

Kellis et al. 2020d. Sequence Alignment and Dynamic Programming (Chapter 2 of Computational Biology: Genomes, Networks, and Evolution). LibreTexts (MIT open textbook). https://bio.libretexts.org/Bookshelves/Computational_Biology/Book%3A_Computational_Biology_-_Genomes_Networks_and_Evolution_(Kellis_et_al.)/02%3A_Sequence_Alignment_and_Dynamic_Programming.

Kellis et al. 2021. Molecular Evolution and Phylogenetics (Chapter 26 of Computational Biology: Genomes, Networks, and Evolution). LibreTexts (MIT open textbook). https://bio.libretexts.org/Bookshelves/Computational_Biology/Book%3A_Computational_Biology_-_Genomes_Networks_and_Evolution_(Kellis_et_al.)/26%3A_Molecular_Evolution_and_Phylogenetics.

Kerfeld, Cheryl A., and Kathleen M. Scott. 2011. “Using BLAST to Teach ‘e-Value-Tionary’ Concepts.” PLoS Biology 9 (2): e1001014. https://doi.org/10.1371/journal.pbio.1001014.

Kiouri, Despoina P., Georgios C. Batsis, and Christos T. Chasapis. 2025. “Structure-Based Approaches for Protein–Protein Interaction Prediction Using Machine Learning and Deep Learning.” Biomolecules 15 (1): 141. https://doi.org/10.3390/biom15010141.

Kobak, Dmitry, and Philipp Berens. 2019. “The Art of Using t-SNE for Single-Cell Transcriptomics.” Nature Communications 10 (1). https://doi.org/10.1038/s41467-019-13056-x.

Kobayashi, Hirofumi, Keith C. Cheveralls, Manuel D. Leonetti, and Loic A. Royer. 2022. “Self-Supervised Deep Learning Encodes High-Resolution Features of Protein Subcellular Localization.” Nature Methods 19 (8): 995–1003. https://doi.org/10.1038/s41592-022-01541-z.

Koboldt, Daniel C. 2020. “Best Practices for Variant Calling in Clinical Sequencing.” Genome Medicine 12 (1). https://doi.org/10.1186/s13073-020-00791-w.

Koonin, Eugene V. 2015. “Origin of Eukaryotes from Within Archaea, Archaeal Eukaryome and Bursts of Gene Gain: Eukaryogenesis Just Made Easier?” Philosophical Transactions of the Royal Society B: Biological Sciences 370 (1678): 20140333. https://doi.org/10.1098/rstb.2014.0333.

Koonin, Eugene V. 2016. “Horizontal Gene Transfer: Essentiality and Evolvability in Prokaryotes, and Roles in Evolutionary Transitions.” F1000Research 5: 1805. https://doi.org/10.12688/f1000research.8737.1.

Korthauer, Keegan, Patrick K. Kimes, Claire Duvallet, et al. 2019. “A Practical Guide to Methods Controlling False Discoveries in Computational Biology.” Genome Biology 20 (1). https://doi.org/10.1186/s13059-019-1716-1.

Kosiol, Carolin, and Maria Anisimova. 2019. “Selection Acting on Genomes.” In Evolutionary Genomics. Springer New York. https://doi.org/10.1007/978-1-4939-9074-0_12.

Koutrouli, Mikaela, Evangelos Karatzas, David Paez-Espino, and Georgios A. Pavlopoulos. 2020. “A Guide to Conquer the Biological Network Era Using Graph Theory.” Frontiers in Bioengineering and Biotechnology 8. https://doi.org/10.3389/fbioe.2020.00034.

Kovalevskiy, Oleg, Juan Mateos-Garcia, and Kathryn Tunyasuvunakool. 2024. AlphaFold Two Years on: Validation and Impact. https://doi.org/10.48550/arXiv.2403.02124.

Kulkarni, Prakash, Lauren Porter, Tsui-Fen Chou, et al. 2025. “Evolving Concepts of the Protein Universe.” iScience 28 (3): 112012. https://doi.org/10.1016/j.isci.2025.112012.

Lapidus, Alla L., and Anton I. Korobeynikov. 2021. “Metagenomic Data Assembly – the Way of Decoding Unknown Microorganisms.” Frontiers in Microbiology 12. https://doi.org/10.3389/fmicb.2021.613791.

Laurents, Douglas V. 2022. “AlphaFold 2 and NMR Spectroscopy: Partners to Understand Protein Structure, Dynamics and Function.” Frontiers in Molecular Biosciences 9. https://doi.org/10.3389/fmolb.2022.906437.

Law, Charity W., Monther Alhamdoosh, Shian Su, et al. 2018. “RNA-Seq Analysis Is Easy as 1-2-3 with Limma, Glimma and edgeR.” F1000Research 5: 1408. https://doi.org/10.12688/f1000research.9005.3.

Lee, Benjamin D., Anthony Gitter, Casey S. Greene, et al. 2022. “Ten Quick Tips for Deep Learning in Biology.” PLOS Computational Biology 18 (3): e1009803. https://doi.org/10.1371/journal.pcbi.1009803.

Li, Qing, Zhihang Hu, Yixuan Wang, et al. 2024. “Progress and Opportunities of Foundation Models in Bioinformatics.” Briefings in Bioinformatics 25 (6). https://doi.org/10.1093/bib/bbae548.

Lin, Baohui, Xiaoling Luo, Yumeng Liu, and Xiaopeng Jin. 2024. “A Comprehensive Review and Comparison of Existing Computational Methods for Protein Function Prediction.” Briefings in Bioinformatics 25 (4). https://doi.org/10.1093/bib/bbae289.

Lin, Pei-Yu, Shiang-Chin Huang, Kuan-Lin Chen, et al. 2025. “Analysing Protein Complexes in Plant Science: Insights and Limitation with AlphaFold 3.” Botanical Studies 66 (1). https://doi.org/10.1186/s40529-025-00462-2.

Liu, Dang, Martin Hunt, and Isheng J Tsai. 2018. “Inferring Synteny Between Genome Assemblies: A Systematic Evaluation.” BMC Bioinformatics 19 (1). https://doi.org/10.1186/s12859-018-2026-4.

Liu, Liang, Christian Anderson, Dennis Pearl, and Scott V. Edwards. 2019. “Modern Phylogenomics: Building Phylogenetic Trees Using the Multispecies Coalescent Model.” In Evolutionary Genomics. Springer New York. https://doi.org/10.1007/978-1-4939-9074-0_7.

Livesey, Benjamin J, and Joseph A Marsh. 2023. “Updated Benchmarking of Variant Effect Predictors Using Deep Mutational Scanning.” Molecular Systems Biology 19 (8). https://doi.org/10.15252/msb.202211474.

Lopez, Romain, Adam Gayoso, and Nir Yosef. 2020. “Enhancing Scientific Discoveries in Molecular Biology with Deep Generative Models.” Molecular Systems Biology 16 (9). https://doi.org/10.15252/msb.20199198.

Love, Michael I., Simon Anders, Vladislav Kim, and Wolfgang Huber. 2016. “RNA-Seq Workflow: Gene-Level Exploratory Analysis and Differential Expression.” F1000Research 4: 1070. https://doi.org/10.12688/f1000research.7035.2.

Lowe, Rohan, Neil Shirley, Mark Bleackley, Stephen Dolan, and Thomas Shafee. 2017. “Transcriptomics Technologies.” PLOS Computational Biology 13 (5): e1005457. https://doi.org/10.1371/journal.pcbi.1005457.

Lubiana, Tiago, Rafael Lopes, Pedro Medeiros, et al. 2023. “Ten Quick Tips for Harnessing the Power of ChatGPT in Computational Biology.” PLOS Computational Biology 19 (8): e1011319. https://doi.org/10.1371/journal.pcbi.1011319.

Lussi, Yvonne C., Michele Magrane, Maria J. Martin, and Sandra and Orchard. 2023. “Searching and Navigating UniProt Databases.” Current Protocols 3 (3). https://doi.org/10.1002/cpz1.700.

Ma, Shaoqian, and Yongyou Zhang. 2020. “Profiling Chromatin Regulatory Landscape: Insights into the Development of ChIP-Seq and ATAC-Seq.” Molecular Biomedicine 1 (1). https://doi.org/10.1186/s43556-020-00009-w.

Makałowski, Wojciech, Valer Gotea, Amit Pande, and Izabela Makałowska. 2019. “Transposable Elements: Classification, Identification, and Their Use as a Tool for Comparative Genomics.” In Evolutionary Genomics. Springer New York. https://doi.org/10.1007/978-1-4939-9074-0_6.

Marchese, Francesco P., Ivan Raimondi, and Maite Huarte. 2017. “The Multidimensional Mechanisms of Long Noncoding RNA Function.” Genome Biology 18 (1). https://doi.org/10.1186/s13059-017-1348-2.

Mardikoraem, Mehrsa, Zirui Wang, Nathaniel Pascual, and Daniel Woldring. 2023. “Generative Models for Protein Sequence Modeling: Recent Advances and Future Directions.” Briefings in Bioinformatics 24 (6). https://doi.org/10.1093/bib/bbad358.

Marku, Malvina, and Vera Pancaldi. 2023. “From Time-Series Transcriptomics to Gene Regulatory Networks: A Review on Inference Methods.” PLOS Computational Biology 19 (8): e1011254. https://doi.org/10.1371/journal.pcbi.1011254.

Masson, Glenn R., John E. Burke, Natalie G. Ahn, et al. 2019. “Recommendations for Performing, Interpreting and Reporting Hydrogen Deuterium Exchange Mass Spectrometry (HDX-MS) Experiments.” Nature Methods 16 (7): 595–602. https://doi.org/10.1038/s41592-019-0459-y.

McHugh, Leo, and Jonathan W Arthur. 2008. “Computational Methods for Protein Identification from Mass Spectrometry Data.” PLoS Computational Biology 4 (2): e12. https://doi.org/10.1371/journal.pcbi.0040012.

Medini, Duccio, Claudio Donati, Rino Rappuoli, and Hervé Tettelin. 2020. “The Pangenome: A Data-Driven Discovery in Biology.” In The Pangenome. Springer International Publishing. https://doi.org/10.1007/978-3-030-38281-0_1.

Miller, Catriona, Theo Portlock, Denis M. Nyaga, and Justin M. O’Sullivan. 2024. “A Review of Model Evaluation Metrics for Machine Learning in Genetics and Genomics.” Frontiers in Bioinformatics 4. https://doi.org/10.3389/fbinf.2024.1457619.

Moore, Josh, Daniela Basurto-Lozada, Sébastien Besson, et al. 2023. “OME-Zarr: A Cloud-Optimized Bioimaging File Format with International Community Support.” Histochemistry and Cell Biology 160 (3): 223–51. https://doi.org/10.1007/s00418-023-02209-1.

Navin, Nicholas E. 2015. “The First Five Years of Single-Cell Cancer Genomics and Beyond.” Genome Research 25 (10): 1499–507. https://doi.org/10.1101/gr.191098.115.

Nguyen, Lan Huong, and Susan Holmes. 2019. “Ten Quick Tips for Effective Dimensionality Reduction.” PLOS Computational Biology 15 (6): e1006907. https://doi.org/10.1371/journal.pcbi.1006907.

Noble, William Stafford, and Michael J. MacCoss. 2012. “Computational and Statistical Analysis of Protein Mass Spectrometry Data.” PLoS Computational Biology 8 (1): e1002296. https://doi.org/10.1371/journal.pcbi.1002296.

Notredame, Cédric. 2007. “Recent Evolutions of Multiple Sequence Alignment Algorithms.” PLoS Computational Biology 3 (8): e123. https://doi.org/10.1371/journal.pcbi.0030123.

Ofer, Dan, Nadav Brandes, and Michal Linial. 2021. “The Language of Proteins: NLP, Machine Learning & Protein Sequences.” Computational and Structural Biotechnology Journal 19: 1750–58. https://doi.org/10.1016/j.csbj.2021.03.022.

Oliveira, Saulo de, and Charlotte Deane. 2017. “Co-Evolution Techniques Are Reshaping the Way We Do Structural Bioinformatics.” F1000Research 6: 1224. https://doi.org/10.12688/f1000research.11543.1.

Oliver, Stephen G., Antonia Lock, Midori A. Harris, Paul Nurse, and Valerie Wood. 2016. “Model Organism Databases: Essential Resources That Need the Support of Both Funders and Users.” BMC Biology 14 (1). https://doi.org/10.1186/s12915-016-0276-z.

Oza, Vishal H., Jordan H. Whitlock, Elizabeth J. Wilk, et al. 2023. “Ten Simple Rules for Using Public Biological Data for Your Research.” PLOS Computational Biology 19 (1): e1010749. https://doi.org/10.1371/journal.pcbi.1010749.

Paggi, Joseph M., Ayush Pandit, and Ron O. Dror. 2024. “The Art and Science of Molecular Docking.” Annual Review of Biochemistry 93 (1): 389–410. https://doi.org/10.1146/annurev-biochem-030222-120000.

Pal, Koustav, Mattia Forcato, and Francesco Ferrari. 2018. “Hi-c Analysis: From Data Generation to Integration.” Biophysical Reviews 11 (1): 67–78. https://doi.org/10.1007/s12551-018-0489-1.

Palazzo, Alexander F., and T. Ryan Gregory. 2014. “The Case for Junk DNA.” PLoS Genetics 10 (5): e1004351. https://doi.org/10.1371/journal.pgen.1004351.

Pan, Xingjie, and Tanja Kortemme. 2021. “Recent Advances in de Novo Protein Design: Principles, Methods, and Applications.” Journal of Biological Chemistry 296: 100558. https://doi.org/10.1016/j.jbc.2021.100558.

Panneerselvam, Kalpana, Pablo Porras, Noemí del-Toro, et al. 2024. “IntAct Database for Accessing IMEx’s Contextual Metadata of Molecular Interactions.” Current Protocols 4 (10). https://doi.org/10.1002/cpz1.70018.

Park, Jiwoon, Junbum Kim, Tyler Lewy, et al. 2022. “Spatial Omics Technologies at Multimodal and Single Cell/Subcellular Level.” Genome Biology 23 (1). https://doi.org/10.1186/s13059-022-02824-6.

Pellicer, Jaume, Oriane Hidalgo, Steven Dodsworth, and Ilia Leitch. 2018. “Genome Size Diversity and Its Impact on the Evolution of Land Plants.” Genes 9 (2): 88. https://doi.org/10.3390/genes9020088.

Pérez-Cobas, Ana Elena, Laura Gomez-Valero, and Carmen Buchrieser. 2020. “Metagenomic Approaches in Microbial Ecology: An Update on Whole-Genome and Marker Gene Sequencing Analyses.” Microbial Genomics 6 (8). https://doi.org/10.1099/mgen.0.000409.

Perez-Riverol, Yasset, Emanuele Alpi, Rui Wang, Henning Hermjakob, and Juan Antonio Vizcaíno. 2015. “Making Proteomics Data Accessible and Reusable: Current State of Proteomics Databases and Repositories.” PROTEOMICS 15 (5-6): 930–50. https://doi.org/10.1002/pmic.201400302.

Perez-Riverol, Yasset, Andrey Zorin, Gaurhari Dass, et al. 2019. “Quantifying the Impact of Public Omics Data.” Nature Communications 10 (1). https://doi.org/10.1038/s41467-019-11461-w.

Phillips, Ben P., and Elizabeth A. Miller. 2021. “Membrane Protein Folding and Quality Control.” Current Opinion in Structural Biology 69: 50–54. https://doi.org/10.1016/j.sbi.2021.03.003.

Ponting, Chris P., and Wilfried Haerty. 2022. “Genome-Wide Analysis of Human Long Noncoding RNAs: A Provocative Review.” Annual Review of Genomics and Human Genetics 23 (1): 153–72. https://doi.org/10.1146/annurev-genom-112921-123710.

Preiner, Martina, Silke Asche, Sidney Becker, et al. 2020. “The Future of Origin of Life Research: Bridging Decades-Old Divisions.” Life 10 (3): 20. https://doi.org/10.3390/life10030020.

Pross, Addy, and Robert Pascal. 2013. “The Origin of Life: What We Know, What We Can Know and What We Will Never Know.” Open Biology 3 (3): 120190. https://doi.org/10.1098/rsob.120190.

Puigbò, Pere, Yuri I. Wolf, and Eugene V. Koonin. 2019. “Genome-Wide Comparative Analysis of Phylogenetic Trees: The Prokaryotic Forest of Life.” In Evolutionary Genomics. Springer New York. https://doi.org/10.1007/978-1-4939-9074-0_8.

Regev, Aviv, Sarah A Teichmann, Eric S Lander, et al. 2017. “The Human Cell Atlas.” eLife 6. https://doi.org/10.7554/elife.27041.

Reis, Mario dos, and Ziheng Yang. 2019. “Bayesian Molecular Clock Dating Using Genome-Scale Datasets.” In Evolutionary Genomics. Springer New York. https://doi.org/10.1007/978-1-4939-9074-0_10.

Rennie, Martin Luke, and Michael R. Oliver. 2025. “Emerging Frontiers in Protein Structure Prediction Following the AlphaFold Revolution.” Journal of The Royal Society Interface 22 (225). https://doi.org/10.1098/rsif.2024.0886.

Riccio, Cristian, Max L. Jansen, Linlin Guo, and Andreas Ziegler. 2024. “Variant Effect Predictors: A Systematic Review and Practical Guide.” Human Genetics 143 (5): 625–34. https://doi.org/10.1007/s00439-024-02670-5.

Richards, Alicia L, Manon Eckhardt, and Nevan J Krogan. 2021. “Mass Spectrometry-Based Protein–Protein Interaction Networks for the Study of Human Diseases.” Molecular Systems Biology 17 (1). https://doi.org/10.15252/msb.20188792.

Rosonovski, Summer, Maria Levchenko, Michele Ide-Smith, Lynne Faulk, Melissa Harrison, and Johanna McEntyre. 2023. “Searching and Evaluating Publications and Preprints Using Europe PMC.” Current Protocols 3 (3). https://doi.org/10.1002/cpz1.694.

Rothfels, Karen, Marija Milacic, Lisa Matthews, et al. 2023. “Using the Reactome Database.” Current Protocols 3 (4). https://doi.org/10.1002/cpz1.722.

Sahakyan, Harutyun, Pascal Mutz, Victor Tobiasson, and Eugene V. Koonin. 2025. “Exploring the Protein Universe with Distant Similarity Detection Methods.” Protein Science 35 (1). https://doi.org/10.1002/pro.70397.

Salzberg, Steven L. 2019. “Next-Generation Genome Annotation: We Still Struggle to Get It Right.” Genome Biology 20 (1). https://doi.org/10.1186/s13059-019-1715-2.

Satam, Heena, Kandarp Joshi, Upasana Mangrolia, et al. 2023. “Next-Generation Sequencing Technology: Current Trends and Advancements.” Biology 12 (7): 997. https://doi.org/10.3390/biology12070997.

Savage, Sara R., and Bing Zhang. 2020. “Using Phosphoproteomics Data to Understand Cellular Signaling: A Comprehensive Guide to Bioinformatics Resources.” Clinical Proteomics 17 (1). https://doi.org/10.1186/s12014-020-09290-x.

Schleper, Christa, and Thiago Rodrigues-Oliveira. 2026. “Asgard Archaea: Have We Found Our Microbial Ancestors?” The EMBO Journal 45 (6): 1836–51. https://doi.org/10.1038/s44318-026-00719-x.

Škunca, Nives, Richard J. Roberts, and Martin Steffen. 2016. “Evaluating Computational Gene Ontology Annotations.” In The Gene Ontology Handbook. Springer New York. https://doi.org/10.1007/978-1-4939-3743-1_8.

Slonim, Donna K., and Itai Yanai. 2009. “Getting Started in Gene Expression Microarray Analysis.” PLoS Computational Biology 5 (10): e1000543. https://doi.org/10.1371/journal.pcbi.1000543.

Smith, Gabriel Reuben, Carolina Bello, Lalasia Bialic-Murphy, et al. 2024. “Ten Simple Rules for Using Large Language Models in Science, Version 1.0.” PLOS Computational Biology 20 (1): e1011767. https://doi.org/10.1371/journal.pcbi.1011767.

Snider, Jamie, Max Kotlyar, Punit Saraon, Zhong Yao, Igor Jurisica, and Igor Stagljar. 2015. “Fundamentals of Protein Interaction Network Mapping.” Molecular Systems Biology 11 (12). https://doi.org/10.15252/msb.20156351.

Sonawane, Abhijeet R., Scott T. Weiss, Kimberly Glass, and Amitabh Sharma. 2019. “Network Medicine in the Age of Biomedical Big Data.” Frontiers in Genetics 10. https://doi.org/10.3389/fgene.2019.00294.

Spang, Anja, Tara A Mahendrarajah, Pierre Offre, and Courtney W Stairs. 2022. “Evolving Perspective on the Origin and Diversification of Cellular Life and the Virosphere.” Genome Biology and Evolution 14 (6). https://doi.org/10.1093/gbe/evac034.

Spudich, Giulietta M, and Xosé M Fernández-Suárez. 2010. “Touring Ensembl: A Practical Guide to Genome Browsing.” BMC Genomics 11 (1): 295. https://doi.org/10.1186/1471-2164-11-295.

Stevens, Jan A., Fabian Grünewald, P. A. Marco van Tilburg, et al. 2023. “Molecular Dynamics Simulation of an Entire Cell.” Frontiers in Chemistry 11. https://doi.org/10.3389/fchem.2023.1106495.

Stirling, David R., Madison J. Swain-Bowden, Alice M. Lucas, Anne E. Carpenter, Beth A. Cimini, and Allen Goodman. 2021. “CellProfiler 4: Improvements in Speed, Utility and Usability.” BMC Bioinformatics 22 (1). https://doi.org/10.1186/s12859-021-04344-9.

Strokach, Alexey, and Philip M. Kim. 2022. “Deep Generative Modeling for Protein Design.” Current Opinion in Structural Biology 72: 226–36. https://doi.org/10.1016/j.sbi.2021.11.008.

Subramanian, Indhupriya, Srikant Verma, Shiva Kumar, Abhay Jere, and Krishanpal Anamika. 2020. “Multi-Omics Data Integration, Interpretation, and Its Application.” Bioinformatics and Biology Insights 14: 117793221989905. https://doi.org/10.1177/1177932219899051.

Taylor, Dylan J., Jordan M. Eizenga, Qiuhui Li, et al. 2024. “Beyond the Human Genome Project: The Age of Complete Human Genome Sequences and Pangenome References.” Annual Review of Genomics and Human Genetics 25 (1): 77–104. https://doi.org/10.1146/annurev-genom-021623-081639.

Terwilliger, Thomas C., Dorothee Liebschner, Tristan I. Croll, et al. 2023. “AlphaFold Predictions Are Valuable Hypotheses and Accelerate but Do Not Replace Experimental Structure Determination.” Nature Methods 21 (1): 110–16. https://doi.org/10.1038/s41592-023-02087-4.

Thomas, Paul D. 2016. “The Gene Ontology and the Meaning of Biological Function.” In The Gene Ontology Handbook. Springer New York. https://doi.org/10.1007/978-1-4939-3743-1_2.

Timp, Winston, and Gregory Timp. 2020. “Beyond Mass Spectrometry, the Next Step in Proteomics.” Science Advances 6 (2). https://doi.org/10.1126/sciadv.aax8978.

Togashi, Yuichi, and Holger Flechsig. 2018. “Coarse-Grained Protein Dynamics Studies Using Elastic Network Models.” International Journal of Molecular Sciences 19 (12): 3899. https://doi.org/10.3390/ijms19123899.

Touw, W. G., J. R. Bayjanov, L. Overmars, et al. 2012. “Data Mining in the Life Sciences with Random Forest: A Walk in the Park or Lost in the Jungle?” Briefings in Bioinformatics 14 (3): 315–26. https://doi.org/10.1093/bib/bbs034.

Tsuchiya, Yuko, Yu Yamamori, and Kentaro Tomii. 2022. “Protein–Protein Interaction Prediction Methods: From Docking-Based to AI-Based Approaches.” Biophysical Reviews 14 (6): 1341–48. https://doi.org/10.1007/s12551-022-01032-7.

Valentini, Giorgio, Dario Malchiodi, Jessica Gliozzo, et al. 2023. “The Promises of Large Language Models for Protein Design and Modeling.” Frontiers in Bioinformatics 3. https://doi.org/10.3389/fbinf.2023.1304099.

Van Buren, Samantha L, C Titus Brown, Tomasz Szmatoła, and Carrie J Finno. 2025. “A Comparative Review of Short Genetic Variant Databases Across Humans and Animal Species.” Briefings in Bioinformatics 26 (4). https://doi.org/10.1093/bib/bbaf356.

Van Oss, Stephen Branden, and Anne-Ruxandra Carvunis. 2019. “De Novo Gene Birth.” PLOS Genetics 15 (5): e1008160. https://doi.org/10.1371/journal.pgen.1008160.

Varadi, Mihaly, Stephen Anyango, Sri Devan Appasamy, et al. 2022. “PDBe and PDBe-KB: Providing High-Quality, up-to-Date and Integrated Resources of Macromolecular Structures to Support Basic and Applied Research and Education.” Protein Science 31 (10). https://doi.org/10.1002/pro.4439.

Vendramin, Roberto, Kevin Litchfield, and Charles Swanton. 2021. “Cancer Evolution: Darwin and Beyond.” The EMBO Journal 40 (18). https://doi.org/10.15252/embj.2021108389.

Vénien-Bryan, Catherine, Zhuolun Li, Laurent Vuillard, and Jean Albert Boutin. 2017. “Cryo-Electron Microscopy and x-Ray Crystallography: Complementary Approaches to Structural Biology and Drug Discovery.” Acta Crystallographica Section F Structural Biology Communications 73 (4): 174–83. https://doi.org/10.1107/s2053230x17003740.

Visscher, Peter M., Naomi R. Wray, Qian Zhang, et al. 2017. “10 Years of GWAS Discovery: Biology, Function, and Translation.” The American Journal of Human Genetics 101 (1): 5–22. https://doi.org/10.1016/j.ajhg.2017.06.005.

Vu, Thi Thuy Duong, and Jaehee Jung. 2021. “Protein Function Prediction with Gene Ontology: From Traditional to Deep Learning Models.” PeerJ 9: e12019. https://doi.org/10.7717/peerj.12019.

Vuong, Paton, Michael J. Wise, Andrew S. Whiteley, and Parwinder Kaur. 2022. “Ten Simple Rules for Investigating (Meta)genomic Data from Environmental Ecosystems.” PLOS Computational Biology 18 (12): e1010675. https://doi.org/10.1371/journal.pcbi.1010675.

Waman, Vaishali P., Nicola Bordin, Rachel Alcraft, et al. 2024. “CATH 2024: CATH-AlphaFlow Doubles the Number of Structures in CATH and Reveals Nearly 200 New Folds.” Journal of Molecular Biology 436 (17): 168551. https://doi.org/10.1016/j.jmb.2024.168551.

Wang, Yan, Hang Zhang, Haolin Zhong, and Zhidong Xue. 2021. “Protein Domain Identification Methods and Online Resources.” Computational and Structural Biotechnology Journal 19: 1145–53. https://doi.org/10.1016/j.csbj.2021.01.041.

Wang, Yen-Wen, Jaqueline Hess, Jason C Slot, and Anne Pringle. 2020. “De Novo Gene Birth, Horizontal Gene Transfer, and Gene Duplication as Sources of New Gene Families Associated with the Origin of Symbiosis inAmanita.” Genome Biology and Evolution 12 (11): 2168–82. https://doi.org/10.1093/gbe/evaa193.

Wang, Zichen, Alexander Lachmann, and Avi Ma’ayan. 2018. “Mining Data and Metadata from the Gene Expression Omnibus.” Biophysical Reviews 11 (1): 103–10. https://doi.org/10.1007/s12551-018-0490-8.

Wang, Zicong, Peng-Kai Liu, and Lingjun Li. 2024. “A Tutorial Review of Labeling Methods in Mass Spectrometry-Based Quantitative Proteomics.” ACS Measurement Science Au 4 (4): 315–37. https://doi.org/10.1021/acsmeasuresciau.4c00007.

Weber, Lukas M., Wouter Saelens, Robrecht Cannoodt, et al. 2019. “Essential Guidelines for Computational Method Benchmarking.” Genome Biology 20 (1). https://doi.org/10.1186/s13059-019-1738-8.

Weissenow, Konstantin, and Burkhard Rost. 2025. “Are Protein Language Models the New Universal Key?” Current Opinion in Structural Biology 91: 102997. https://doi.org/10.1016/j.sbi.2025.102997.

Wen, Bo, Jack Freestone, Michael Riffle, Michael J. MacCoss, William S. Noble, and Uri Keich. 2025. “Assessment of False Discovery Rate Control in Tandem Mass Spectrometry Analysis Using Entrapment.” Nature Methods 22 (7): 1454–63. https://doi.org/10.1038/s41592-025-02719-x.

Wen, Bo, Wen-Feng Zeng, Yuxing Liao, et al. 2020. “Deep Learning in Proteomics.” PROTEOMICS 20 (21-22). https://doi.org/10.1002/pmic.201900335.

Whitley, Paul, Brayan Grau, James C. Gumbart, Luis Martínez-Gil, and Ismael Mingarro. 2021. “Folding and Insertion of Transmembrane Helices at the ER.” International Journal of Molecular Sciences 22 (23): 12778. https://doi.org/10.3390/ijms222312778.

Wijesooriya, Kaumadi, Sameer A. Jadaan, Kaushalya L. Perera, Tanuveer Kaur, and Mark Ziemann. 2022. “Urgent Need for Consistent Standards in Functional Enrichment Analysis.” PLOS Computational Biology 18 (3): e1009935. https://doi.org/10.1371/journal.pcbi.1009935.

Williams, Cameron G., Hyun Jae Lee, Takahiro Asatsuma, Roser Vento-Tormo, and Ashraful Haque. 2022. “An Introduction to Spatial Transcriptomics for Biomedical Research.” Genome Medicine 14 (1). https://doi.org/10.1186/s13073-022-01075-1.

Winnifrith, Adam, Carlos Outeiral, and Brian L. Hie. 2024. “Generative Artificial Intelligence for de Novo Protein Design.” Current Opinion in Structural Biology 86: 102794. https://doi.org/10.1016/j.sbi.2024.102794.

Wolf, Aaron B., and Joshua M. Akey. 2018. “Outstanding Questions in the Study of Archaic Hominin Admixture.” PLOS Genetics 14 (5): e1007349. https://doi.org/10.1371/journal.pgen.1007349.

Wooley, John C., Adam Godzik, and Iddo Friedberg. 2010. “A Primer on Metagenomics.” PLoS Computational Biology 6 (2): e1000667. https://doi.org/10.1371/journal.pcbi.1000667.

Xiao, Hanyu, Yijin Zou, Jieqiong Wang, and Shibiao Wan. 2024. “A Review for Artificial Intelligence Based Protein Subcellular Localization.” Biomolecules 14 (4): 409. https://doi.org/10.3390/biom14040409.

Yoon, Byung-Jun. 2009. “Hidden Markov Models and Their Applications in Biological Sequence Analysis.” Current Genomics 10 (6): 402–15. https://doi.org/10.2174/138920209789177575.

Young, Lindsey N., and Elizabeth Villa. 2023. “Bringing Structure to Cell Biology with Cryo-Electron Tomography.” Annual Review of Biophysics 52 (1): 573–95. https://doi.org/10.1146/annurev-biophys-111622-091327.

Youssef, Noor, Aidan Budd, and Joseph P. Bielawski. 2019. “Introduction to Genome Biology and Diversity.” In Evolutionary Genomics. Springer New York. https://doi.org/10.1007/978-1-4939-9074-0_1.

Zhang, Shuang, Rui Fan, Yuti Liu, Shuang Chen, Qiao Liu, and Wanwen Zeng. 2023. “Applications of Transformer-Based Language Models in Bioinformatics: A Survey.” Bioinformatics Advances 3 (1). https://doi.org/10.1093/bioadv/vbad001.

Zhang, Zhidian, Hannah K. Wayment-Steele, Garyk Brixi, Haobo Wang, Dorothee Kern, and Sergey Ovchinnikov. 2024. “Protein Language Models Learn Evolutionary Statistics of Interacting Sequence Motifs.” Proceedings of the National Academy of Sciences 121 (45). https://doi.org/10.1073/pnas.2406285121.

Zhao, Qi, Zheng Zhao, Xiaoya Fan, Zhengwei Yuan, Qian Mao, and Yudong Yao. 2021. “Review of Machine Learning Methods for RNA Secondary Structure Prediction.” PLOS Computational Biology 17 (8): e1009291. https://doi.org/10.1371/journal.pcbi.1009291.

Zielezinski, Andrzej, Hani Z. Girgis, Guillaume Bernard, et al. 2019. “Benchmarking of Alignment-Free Sequence Comparison Methods.” Genome Biology 20 (1). https://doi.org/10.1186/s13059-019-1755-7.

Zverinova, Stepanka, and Victor Guryev. 2021. “Variant Calling: Considerations, Practices, and Developments.” Human Mutation 43 (8): 976–85. https://doi.org/10.1002/humu.24311.