Protein-protein interactions are keys to executing important cellular functions. We hypothesize that gain or loss of protein-protein interactions plays important roles in carcinogenesis. We propose a network-based approach to biomarker discovery that uses protein-protein interactions and regulatory transcriptional networks. We will use prostate cancer and gliomas as models. Prostate cancer is the most common form of cancer and the leading cause of cancer death among men in the developed countries. Glioblastoma multiforme (GBM) is the most common and most aggressive type of primary brain tumor and it GBM accounts for 52% of all primary brain tumor cases. Novel markers and therapeutic targets are still needed for these two cancers. We will be focusing on the androgen receptor protein-protein network in prostate cancer and TGF-beta mediated network in gliomas. Protein interactions will be extracted by text mining, from databases, and from structural data on protein complexes. To identify transcription factors and their targets in human, ChIP-Chip and ChIP-Seq experiments will be carried out. The resulting networks serve as a scaffold on which data of deregulated proteins derived from microarray and next-generation sequencing of patient cancer samples will be mapped. The perturbed subnetwork will be visualized with a new method that identifies and highlights network motifs and modules. Promising biomarker candidates will be further examined and validated experimentally. Where possible, candidates will be modelled using protein structural data on protein-protein complexes, providing the basis to design ligands that occupy binding sites in order to disrupt cancer-relevant interactions. Our proposal is highly innovative. We expect to identify key nodal points in the network, to which protein-protein interactions can be disrupted as a way to perturb or interfere with the network. These key nodal points will serve both as biomarkers and therapeutic targets.