A summary of PANDA gene regulatory network reconstruction and the identification of subtype-specific subnetworks. (A) We combine transcription factor motif and physical protein-protein interaction (PPI) data with gene expression data from each subtype to build individual network models. (B) We compare the weights of edges predicted by PANDA for each of the network models. Each point in the graph represents an edge connecting a TF to a target gene. We also define subnetworks by selecting high-probability edges specific to either the angiogenic (red) and non-angiogenic (blue) model. The number of edges and genes identified as part of these subnetworks is noted. (C) Although the subnetworks contain unique sets of edges, genes targeted by TFs in these interactions are not necessarily unique, suggesting distinct regulatory processes.