Co-clustering of diseases, genes, and drugs for identification of their related gene modules

Abstract

Finding gene clusters that can be shared between drugs and diseases plays an important role in drug discovery. Targeting disease causing genes directly in drug development can increase the chance of drug approval through the clinical phase. This paper introduces a new co-clustering approach on the tripartite graph of genes, drugs, and diseases. As a result of co-clustering, gene modules and their related drugs and diseases are identified. It is shown that identified gene modules are functionally related. In addition the resulted gene modules are closely connected to each other in the protein-protein interaction network compared to that of random gene selection. The resulting gene modules can be used for investigating the genes that can be targeted with new drugs for treatment of diseases that are co-clustered with them. The proposed method is scalable and can be used for other multi-view graph co-clustering applications like social networks.