Centrality Combination Method Based on Feature Selection for Protein Interaction Networks

Essential proteins are important participants in various life activities and play a vital role the survival reproduction of life. The network-based centrality methods common way to identify essential for protein interaction networks. Due differences between existing methods, it is feasible approach improve identification accuracy by combining methods. In this paper, we propose combination method based on feature selection. First, measure values 14 classical viewed as data. Then, subset relevant features selected according importance features. Finally, corresponding combined using geometric mean proteins. To verify effectiveness method, apply original static network (SPIN), dynamic (DPIN) refined (RDPIN), compare result with those each single (LAC, DC, DMNC, NC, TP, CLC, BC, LC, CC, KC, CR, EC, PR, LR). experimental results shows that achieves better prediction performance than mehtods terms precision, sensitivity, specificity, positive predictive value, negative F-measure rate. It has been illustrated proposed can help more accurately.