Figure 1.

Overview of the proposed MultiFDRnet method. (a) Constructing a multiplex network using multiple PPI networks. (b) Estimating local FDR scores through an empirical Bayes analysis. Red vertical bars are estimated counts of non-null genes. (c) Performing random walk to quantify subnetworks. Red nodes represent two state nodes corresponding to a seed gene and blue arrows represent the possible random walks originating from one state node. (d) Detecting significantly perturbed subnetworks for given seeds by solving mixed-integer linear programming problems. An example solution is the subnetwork within the red circles.