THE DEPENDENCE OF AMINO-ACID PAIR CORRELATIONS ON STRUCTURAL ENVIRONMENT

A statistical analysis was performed to determine to what extent an am ino acid determines the identity of its neighbors and to what extent t his is determined by the structural environment. Log-Linear analysis w as used to discriminate chance occurrence from statistically meaningfu l correlations, The classification of structures was arbitrary, but wa s also tested for significance, A list of statistically significant in teraction types was selected and then ranked according to apparent imp ortance for applications such as protein design. This showed that, in general, nonlocal, through-space interactions were more important than those between residues near in the protein sequence. The highest rank ed nonlocal interactions involved residues in beta-sheet structures. O f the local interactions, those between residues i and i + 2 were the most important in both alpha-helices and beta-strands, Some surprising ly strong correlations were discovered within beta-sheets between resi dues and sites sequentially near to their bridging partners. The resul ts have a clear bearing on protein engineering studies, but also have implications for the construction of knowledge-based force fields. Pro teins 32:175-189, 1998. (C) 1998 Wiley-Liss, Inc.