Authors: Jun Ren; Jianxin Wang; Min Li; Fangxiang Wu
Addresses: School of Information Science and Engineering, Central South University, Changsha 410083, China; College of Information Science and Technology, Hunan Agricultural University, Changsha 410128, China ' School of Information Science and Engineering, Central South University, Changsha 410083, China ' School of Information Science and Engineering, Central South University, Changsha 410083, China ' Department of Mechanical Engineering and Division of Biomedical Engineering, University of Saskatchewan, Saskatoon, SK S7N 5A9, Canada
Abstract: Most computational methods for identifying essential proteins focus on the topological centrality of protein-protein interaction (PPI) networks. However, these methods have limitations, such as the difficulty for identifying essential proteins with low centrality values and the poor performance for incomplete PPI network. In this paper, protein complex is proven to be an important factor for determining protein essentiality and a new centrality measure, complex centrality, is proposed. The weighted average of complex centrality and subgraph centrality, called harmonic centrality (HC), is proposed to predict essential proteins. It combines PPI network topology and protein complex information and has better performance than methods based on PPI network. The improvement is higher when the PPI network is incomplete. Furthermore, a weighted PPI network is generated by integrating cellular localisation and biological process to a PPI network. The performance of HC measure is improved 5% in this weighted PPI network.
Keywords: centrality measures; complex centrality; essential proteins; harmonic centrality; PPI networks; protein-protein interaction; protein complex; bioinformatics.
International Journal of Data Mining and Bioinformatics, 2015 Vol.12 No.1, pp.24 - 43
Received: 02 May 2012
Accepted: 03 Jun 2013
Published online: 22 Apr 2015 *