Prediction of Essential Proteins by Integration of PPI Network Topology and Protein Complexes Information

Abstract

Identifying essential proteins is important for understanding the minimal requirements for cellular survival and development. Numerous computational methods have been proposed to identify essential proteins from protein-protein interaction (PPI) network. However most of methods only use the PPI network topology information. HartGT indicated that essentiality is a product of the protein complex rather than the individual protein. Based on these, we propose a new method ECC to identify essential proteins by integration of subgraph centrality (SC) of PPI network and protein complexes information. We apply ECC method and six centrality methods on the yeast PPI network. The experimental results show that the performance of ECC is much better than that of six centrality methods, which means that the prediction of essential proteins based on both network topology and protein complexes set is much better than that only based on network topology. Moreover, ECC has a significant improvement in prediction of low-connectivity essential proteins.