A Core-Attach Based Method for Identifying Protein Complexes in Dynamic PPI Networks

Abstract

Indentifying protein complexes is essential to understanding the principles of cellular systems. Many computational methods have been developed to identify protein complexes in static protein-protein interaction (PPI) network. However, PPI network changes over time, the important dynamics within PPI network is overlooked by these methods. Therefore, discovering complexes in dynamic PPI networks (DPN) is important. DPN contains a series of time-sequenced subnetworks which represent PPI at different time points of cell cycle. In this paper, we propose a dynamic core-attachment algorithm (DCA) to discover protein complexes in DPN. Based on core-attachment assumption, we first detect cores which are small, dense subgraphs and frequently active in the DPN, and then we form complexes by adding short-lived attachments to cores. We apply our DCA to the data of S.cerevisiae and the experimental result shows that DCA outperforms six other complex discovery algorithms, moreover, it reveals that our DCA not only provides dynamic information but also discovers more accurate protein complexes.