Matthew Ruffalo, Ziv Bar-Joseph
We introduce a method that allows for predicting the disruption of specific interactions in cancer patients using somatic mutation data and condition independent protein interaction networks. We apply this method to mutation data from TCGA breast cancer samples, and show that our predictions of edge disruption show significant association with patient survival, with known ligand binding site mutations, and with simulations of protein binding disruption. In addition to agreeing with disruptions of protein interactions from known ligand binding site mutations, this method also shows promise in identifying protein interactions which were previously not known to be affected by somatic mutations.