Network analysis of transmembrane protein structures

Abstract

Most studies have shown that globular proteins exist in small-world networks. The present study is an attempt to determine differences in network parameters between transmembrane and globular proteins. Each protein is represented as an undirected graph, where residues represent nodes and inter-residue interactions as the edges. This was then compared to the degree-preserved random controls, to observe if any variation existed. Results indicate that there is a significant difference in shortest path lengths between transmembrane and globular proteins. Hydrophobic amino acids were found to be more spatially distributed in the transmembrane than in globular proteins causing significantly higher values of shortest path lengths (L). Assortativity values too, were found to be significantly higher in the transmembrane than in globular proteins which is due to the highly connected amino acid residues being exposed to the solvent in transmembrane proteins. On analyzing the closeness centrality, it was found that globular proteins have significantly higher values than the transmembrane proteins. We therefore conclude that transmembrane proteins possess small-world characteristics similar to that of globular proteins.