Fig. 5From: ClusterM: a scalable algorithm for computational prediction of conserved protein complexes across multiple protein interaction networksIllustration of the ClusterM procedure. The red dashed line circles the protein spine across these two networks. At first, for each network, based on the personalized PageRank vector starting from the proteins in the protein spine, we select a group of proteins marked in blue for each network. Then we refine the results by mixed integer programming and obtain the subnetworks in the gray shades, which are well separated from the rest of the networks. With them, we integrate the topological information of the selected subnetworks and the homology information between the proteins in the subnetworks, which are represented by the black dashed lines, and identify the conserved protein complex based on the cost function (6). The proteins in the golden color are proteins in the conserved protein complex identified by ClusterMBack to article page