Protein sequence divergence relative to protein folding

This dissertation reports studies on protein sequence divergence relative to protein folding. A protein database was created containing multiple sequence alignments of homologous families and their corresponding secondary structure information. This database was queried to learn how the sequences of natural proteins might be analyzed to predict parses, which are breaks in standard secondary structural units (helices and strands) in the folded form. Specific strings of consecutive amino acids in a polypeptide chain were identified that improved assignments of parses based on single-residue analysis.; Aspects of protein conformation were analyzed by assigning interior and surface residues from patterns of variation and conservation in homologous protein sequences. This study improved the ability to predict secondary structural units.; These heuristics are described in detail and their performance evaluated when applied to ten protein families with known three-dimensional structures. The use of these heuristics is discussed in the context of protein secondary structure prediction.