Biology through network lenses
Thursday, March 1, 2012 - 5:00pm - 5:15pm
Prediction of protein function and the role of protein pathways in disease from protein-protein interaction (PPI) networks have received much attention in the post-genomic era. Given a limited functional knowledge about some proteins in a network, the challenge of computational and systems biology is to functionally characterize other proteins based on their network connectivity information. For this, an efficient method for capturing proteins' complex interaction patterns in the network is needed. We devise such a mathematically rigorous method that summarizes complex wirings around a protein in the network and compares the topological similarity of proteins' extended network neighborhoods. We use this method to predict protein function and imply involvement of genes (i.e., their protein products) in cancer. Specifically, using many clustering approaches, we group together topologically similar proteins in the human PPI network. The resulting clusters are enriched in biological function: the more topologically similar the extended network neighborhoods of proteins, the more likely the proteins are to perform a same biological function. The resulting clusters are also enriched in cancer genes: cancer and non-cancer genes have different network topological signatures. Also, we address an open debate about whether genes involved in key biological processes exhibit some topological centrality compared to the rest of the proteins in the human PPI network, since changes in their activity are likely to affect the activity of many other proteins. Indeed, we find that genes involved in aging, cancer, infectious diseases, or signaling and drug-targeted pathways occupy topologically complex and dense regions of the network and correspond to its spine that dominates other genes in the network, i.e., connects all other network parts, and can thus pass cellular signals efficiently throughout the network. Hence, network topology is an invaluable source of biological information that can suggest novel drug targets for therapeutics intervention.