Bona fide prediction of aspects of protein conformation. Assigning interior and surface residues from patterns of variation and conservation in homologous protein sequences

Abstract

Heuristics have been developed for analyzing patterns of conservation and variation within a set of aligned homologous protein sequences for the purpose of assigning amino acids whose side-chains lie on the surface and inside the folded structure of a protein. These were used in several recent bona fide predictions of the secondary structure of proteins from sequence data, made and published before crystallographic information became available. Heuristics based on concurrent hydrophilic variation identify positions that lie on the surface. Heuristics based on concurrent hydrophobic conservation and variation identify positions lying in the interior. These heuristics are described here in detail and their performance evaluated when applied to seven protein families with known three-dimensional structures. The performance of individual heuristics is shown to depend on the nature of the multiple alignment within the protein family, and a strategy is presented for obtaining surface and interior assignments useful for predicting secondary structure.