Gene ontology based function prediction of human protein using protein sequence and neighborhood property of PPI network

Abstract

Predicting functions of protein from its amino acid sequence and interacting protein partner is one of the major challenges in post genomic era compared with costly, time consuming biological wet lab techniques. In drug discovery, target protein identification is important step as its inhibition may disturb the activities of pathogen. So, the knowledge of protein function is necessary to inspect the cause of diseases. In this work, we have proposed two function prediction methods FunPred1.1 and FunPred1.2 which use neighbourhood analysis of unknown protein empowered with Amino Acid physico-chemical properties. The basic objective and working of these two methods are almost similar but FunPred1.1 works on the entire neighbourhood graph of unknown protein whereas FunPred1.2 does same with greater efficiency on the densely connected neighbourhood graph considering edge clustering coefficient. In terms of time and performance, FunPred1.2 achieves better than FunPred1.1. All the relevant data, source code and detailed performance on test data are available for download at FunPred-1.