Fig. 1

Illustration of LPDriver. I Constructing the gene interaction network G₁ and G₂. II Removing the edges that exist in both G₁ and G₂. III Removing the edges(interactions) that are not supported by the known gene interaction network. IV Transform G_p(V,E) into bipartite graph and identify driver genes of patient p in bipartite graph. V Identify the potential driver genes of all patients to construct the matrix M_gp. VI Compute the linear neighborhood similarity of genes among patients in M_gp. VII Adopt label propagation method based on matrix S to refine M_gp. VIII Identify highly rank genes as the personalized driver genes from F*