AN ALL-ATOM DISTANCE-DEPENDENT CONDITIONAL-PROBABILITY DISCRIMINATORYFUNCTION FOR PROTEIN-STRUCTURE PREDICTION

We present a formalism to compute the probability of an amino acid seq uence conformation being native-like, given a set of pairwise atom-ato m distances. The formalism is used to derive three discriminatory func tions with different types of representations for the atom-atom contac ts observed in a database of protein structures. These functions inclu de two virtual atom representations and one all-heavy atom representat ion. When applied to six different decoy sets containing a range of co rrect and incorrect conformations of amino acid sequences, the all-ato m distance-dependent discriminatory function is able to identify corre ct from incorrect more often than the discriminatory functions using a pproximate representations. We illustrate the importance of using a de tailed atomic description for obtaining the most accurate discriminati on, and the necessity for testing discriminatory functions against a w ide variety of decoys. The discriminatory function is also shown to be capable of capturing the fine details of atom-atom preferences. These results suggest that the all-atom distance-dependent discriminatory f unction will be useful for protein structure prediction and model refi nement. (C) 1998 Academic Press Limited.