An Ontology based Method for Identifying Highly Interacted Human Protein Complexes and its Pathways

Abstract

Discovering highly interacting proteins and its pathways is a challenge for computational biologists. Identifying interactions among proteins and its pathways has been found to be useful for drug development. For example, Arabidopsis thaliana has 549 annotated metabolic pathways and a few thousand biological processes as defined with gene ontology terms, but the regulators for most of these pathways have not yet been revealed (1–3). With the advent of the whole-genome approach and the explosion of biological data in public repositories, demands have heightened for computational algorithms that can be used to predict pathway regulators using high-throughput gene expression datasets.

In our study, we collected data of 40,788 protein - protein interactions from HPRD and IntAct. This pooled off data was loaded into cytoscape (version 2.63) to visualize the human interactome network through the grid layout method. By using Connected Component (CC) algorithm, the largest and the highly connected human network were found. From 36,945 binary protein – protein interactions, 89 highly connected modules were selected, which are of high score value. The total numbers of proteins in each of the 89 modules were found out. The proteins from all the 89 clusters were classified into 2 categories - normal and diseased pathways. From all these 89 clusters, 1350 proteins were obtained and were unique, of which 374 proteins are in the normal pathways and 976 proteins in diseased pathways. But 976 proteins were found in both normal and diseased pathways. Further computational studies can help to understand the changes that occur in the proteins to become in the diseased pathways.