NEWS ! Pcons-meta server

Now you can run Pcons from our local meta-server. It offers several advantages over the other servers that use Pcons. Try it yourself.

Pcons a consensus fold recognition predictor

Pcons was the first consensus server for fold recognition. It selects the best prediction out of several predictions. For each query sequence predictions from several fold recognition servers is collected. For each of these models a measure that relates to the quality of the model is calculated. The prediction of this new measure is accomplished by utilizing structural comparisons between the models and analyzing the server score for a particular model. Pcons makes at least 10% more correct predictions than the best single method and the specificity is significantly better.

Different Pcons version

Since the development of Pcons-1 during fall 2000 we have developed 4 new versions. The differences between them will shortly be described below.

Pcons-1: Used 5 different servers and a Neural network to predict the quality. It was trained using LiveBench-1 data and tested in LiveBench-2. Pcons-1 is now canceled.
Pcons-2: Used 7 different servers and a linear regression to predict the quality. Quality was measured using an updated version of LGscore. It was trained using LiveBench-2 data and tested in CASP5, CAFASP3 and LiveBench-6. It is also the bases for Pmodeller.
Pcons-3: Used 3, i.e. fewer, different servers and a simplified method to measure similarity between models. It was trained on LiveBench-4 data. Due to problems with ShotGun models and Pmodeller (see below) Pcons-3 is now canceled.
Pcons-4Due to problems with "ShotGun" models Pcons-3 was replaced by Pcons-4 and uses 4 different servers. It was also trained on LiveBench-4 data.
Pcons-5: Developed after CASP5 and made to be more flexible, i.e. easier to upgrade. It uses a variable number of servers + information from a structural evaluation of the models using a C-alpha based version of ProQ.

Pmodeller vs Pcons

Pmodeller is an addition on top of Pcons. In Pmodeller all atom models are built, and also evaluated using ProQ. It has been shown in CASP5 that Pmodeller works better than Pcons. Currently Pmodeller (based on Pcons-2) and Pmodeller-4 (based on Pcons-4) is used. To use it you need a valid license for MODELLER.

Pmodeller-5 is still being developed as Pcons-5 includes the ProQ evaluation by itself.

Pcons6 and Pmodeller-6 has been tested.

Submission

You can submit a sequence to Pcons using meta servers either at

http://bioinfo.pl Pcons II, IV and V
http://genesilico.pl Pcons II and V
Now we have an official meta-server available at www.cbr.su.se it uses Pcons-V, ProQ and other methods that we developed for CASP7

References

Submit a sequence to Pcons
If you use Pcons please cite:
- Lundström J, Rychlewski L, Bujnicki J, Elofsson A. Pcons: A neural-network-based consensus predictor that improves fold recognition. Protein Sci. 2001 Nov;10(11):2354-62 pdf
- "Automatic consensus based fold recognition using Pcons, ProQ and Pmodeller." Björn Wallner, Huisheng Fang and Arne Elofsson, Proteins 2003 53(S6):534-541 pdf
- Björn Wallner and Arne Elofsson "Pcons5: combining consensus, structural evaluation and fold recognition scores", Bioinformatics 2005 21(23):4248-54. pdf
Here is a Paper describing Pcons 2
LiveBench a continuous benchmarking of fold recognition methods (including Pcons)

FAQ:

What is a significant Pcons score ?
Any score higher than 1.5 should be significant

Arne Elofsson