On similarity of PPI subnetworks induced by important proteins: A case study

Identification of protein complexes in PPI networks is one of the major challenges in analysis of PPI networks. In this study we analyze the similarity of PPI subnetworks which are induced by important proteins from protein complexes. In the first phase, we determine which proteins frequently appear in protein complexes and which proteins are central in a PPI network. To measure centrality we examine three well-known methods: Degree centrality, Closeness centrality and Betweenness centrality. In the second phase, we induce subnetworks by frequent and central proteins and measure the similarity of the obtained subnetworks by making use of two standard graph - kernel functions: Random Walk and Shortest Path. In the case study, we test our approach on three different PPI networks combined with four different gold standards of protein complexes. Obtained results indicate that there exists certain similarity between considered subnetworks. This conclusion can help for further investigating of PPI networks and protein complexes.