MEETING REVIEW - THE 2ND MEETING ON THE CRITICAL-ASSESSMENT OF TECHNIQUES FOR PROTEIN-STRUCTURE PREDICTION (CASP2), ASILOMAR, CALIFORNIA, DECEMBER 13-16, 1996

In most fields of scientific endeavor, the outcomes of important exper iments are not always known before the experiments are performed. But in protein structure prediction, algorithms are usually developed and tested in situations where the answers are known. In December 1996, th e Second Meeting on the Critical Assessment of Techniques for Protein Structure Prediction (CASP2) was held in Asilomar, California to recti fy this situation: protein sequences were provided in advance for whic h the experimental structure had not yet been published. Over 70 resea rch groups provided bona fide predictions on 42 targets in four catego ries: comparative or 'homology' modeling, fold recognition or 'threadi ng', ab initio structure predictions, and docking predictions. Since t he previous GASP meeting in 1994, the role of fold recognition in stru cture prediction has increased enormously with the largest number of g roups participating in this category. In this review, we highlight som e of the important developments and give at least a qualitative sense of what kind of methods produced some of the better predictions.