Spectral clustering for detecting protein complexes in protein–protein interaction (PPI) networks

In this paper, we study spectral clustering for detecting protein complexes in PPI (protein–protein interaction) networks, focusing on two open issues: (i) constructing similarity graphs; and (ii) determining the number of clusters. First, we study four similarity graphs to construct graph Laplacian matrices. Then we propose a method to determine the number of clusters based on the properties of PPI networks. Experimental results on PPI networks from DIP data and MIPS data indicate that each similarity graph shows its strengths and disadvantages, and our finding of the number of clusters improves the clustering quality. Finally, spectral clustering obtains results in detecting protein complexes that are comparable to those obtained from several other typical algorithms.