Extreme tumor heterogeneity is the main underlying cause for treatment failure in aggressive glioblastoma (GBM) tumors. For the development of more effective therapies, we need to fully understand the regulatory mechanisms at play. Increasing amount of single cell omics analyses allow to dissect complex tissues into specific cell states with their own gene regulatory networks (GRNs). I hypothesize that a spectrum of cell state-specific GRNs determines regulatory heterogeneity and plasticity in GBM and that advanced computational methodologies are needed to elucidate this regulatory heterogeneity. First, I will benchmark several scalable, state-of-the-art network inference methods for patient-specific and cell state-specific GRNs. The best performing methods will subsequently be used in an innovative meta-analysis to integrate single cell omics across tumors to generate a comprehensive spectrum of cell state-specific GRNs. I will also develop an expandable, deep learning framework for the same purpose. Next, I will study the higher order organization of these cell state-specific GRNs towards multicellular programs. Finally, I will validate in silico and in vitro the accuracy of predicted key regulators and pathways from patient- and cell-state specific GRNs in CRISPRi and drugging experiments. In the longer term identified key regulators and druggable network dependencies will contribute to novel therapeutic opportunities for GBM patients.