CAPRI: Assessing protein docking algorithms in the blind structure prediction of protein-protein complexes

COMP 84

Joel Janin, Laboratoire d'Enzymologie et Biochimie Structurales, Laboratoire d'Enzymologie et Biochimie Structurales, UPR9063, CNRS, 91198-Gif-sur-Yvette, France
CAPRI (Critical Assessment of PRedicted Interactions) is a CASP-like experiment to assess protein-protein docking procedures. Predictors are given atomic coordinates for two proteins. They perform a blind prediction of the complex and submit models to the http://capri.ebi.ac.uk Web site run by K. Henrick at the EBI (Hinxton, UK). These models are then assessed by comparison with unpublished X-ray structures of the protein-protein complexes, kindly provided by their authors to the CAPRI Management Group on a confidential basis. In four rounds of CAPRI prediction involving 13 target complexes, some 25 predictor groups have submitted a total of 1629 models. The evaluation procedure was carried out by S.J. Wodak & R. Mendez (Free University of Brussels, Belgium. Models were judged of high quality if they had over 50% of the native contacts and an interface RMSD < 1 ; good if >30% native contacts, RMSD < 2 ; acceptable if >10% native contacts, RMSD < 4 . Of the nine targets of Rounds 1 to 3, five had at least one good model in the submissions, whereas three targets were not predicted at all. Overall, the CAPRI experiment reveals a growing interest in predicting protein-protein interaction. Genuine progress was evident over three years of the experiment. The results of CAPRI point out the need for faster algorithms, better scoring functions and more effective methods to handle conformational flexibility. Incorporating information from biochemical experiments and sequence analysis is another key element of success. CAPRI is a powerful drive for computational biologists who develop docking algorithms. Its continuation entirely depends on the willingness of structural biologists to provide experimental information. The CAPRI Management Group expresses thanks to those who already did, and calls upon all to support the experiment by contributing targets. Reference: Proteins Vol. 52 pp. 1-122, Special Issue July 1, 2003
 

Docking and Scoring
9:00 AM-12:20 PM, Tuesday, August 24, 2004 Pennsylvania Convention Center -- 109B, Oral

Division of Computers in Chemistry

The 228th ACS National Meeting, in Philadelphia, PA, August 22-26, 2004