We propose a novel graph-based optimization approach to modeling gene regulatory networks for the purpose of understanding gene interactions and pathway function in the presence of mutations. We use a maximum flow formulation to evaluate pathway disruption which provides biological insight into the optimal drug targets for a given mutation profile. Empirical results suggest that this approach recovers biologically valid relationships between genes. We extend the work by Bertsimas and Zhuo to incorporate this structured pathway model and demonstrate recovery of new drug-target interactions.