Contact prediction in protein modeling: Scoring, folding and refinement of coarse-grained models-4

E scoring function was based on the contact data set from the best two predictors (Baker and PROFcon). The accuracy of the contact data used for scoring 5 models of each target is plotted against the RMSD of the model ranked as first in CASP6 by the Kolinski-Bujnicki group (green squares) and against the RMSD of the protein model ranked as first by the contact-based scoring function (red lines which join corresponding points). Results for NF and FR/A categories are presented separately. The most significant improvement is observed in the case of FR/A targets with the accuracy of the contact prediction range of 15–30%. In a similar way the results of the refinement simulations are presented in the right-hand panels (b). The refinement simulations performed better than the post-simulation ranking of the models.<p><b>Copyright information:</b></p><p>Taken from "Contact prediction in protein modeling: Scoring, folding and refinement of coarse-grained models"</p><p>http://www.biomedcentral.com/1472-6807/8/36</p><p>BMC Structural Biology 2008;8():36-36.</p><p>Published online 11 Aug 2008</p><p>PMCID:PMC2527566.</p><p>