Identifying influential genes in protein-protein interaction networks

Influential nodes in influence maximization problems are of great importance for the spread of information in complex networks. In this study, we identify influential nodes, called influential genes, in protein-protein interaction (PPI) networks. In theory, information can percolate through an entire network when influential genes are activated. We propose a new framework by taking the asymmetry of influence into account to identify genes that are more influential in PPI networks. In the framework, we identify influential genes by considering the heterogeneity of influence. As such, the minimal set of influential genes in the influence maximization problem can be mapped onto the optimal set of genes in the optimal percolation problem. We identify the influential genes in the PPI networks of five species, and the results show that the genes identified by our method are more influential and tend to be located in the core of a PPI network. In addition, we find that influential genes tend to be more significantly enriched in essential yeast genes, tumor suppressor genes, and drug target genes.