2008 OSU Molecular Life Sciences
Interdisciplinary Graduate Programs Symposium
Poster abstracts
Abstract:
Understanding residue associations and their structural implications in protein family alignments is an important research direction to extract distinctive constraints on the family. Unfortunately, even advanced probabilistic models assume independence of positions in family alignments while building family descriptions. In this study, we examined interpositional residue associations that exist in protein family alignments. We analyzed structural implications in the presence of such associations to demonstrate their significance for the family description. The phi coefficient was used to measure the degree of association between every pair of residues in every pair of columns of the family alignments. The statistical significances of residue associations were calculated by Fisher’s Exact Test. We generated all possible pairwise residue associations for 961 protein family alignments of Pfam database that have 20 to 500 sequences and have at least one structural reference. Then, we examined actual physical distances of associated residues and their residence on the secondary structural elements. We observed that actual physical distances between associated residues decrease as the phi coefficient and statistical significance increase. Specifically, associations between aromatic residues and charged residues are present in closer proximity compared to other physicochemical properties. We also observed that associated residues tend to reside on more flexible regions of the protein rather than completely being on helices or beta strands.
Keywords: protein family alignment, correlation, protein structure