Computational Studies Reveal Varied Mechanisms of Oncogenic Activation amidst a Varied Mutational Landscape in Cancer Patients

Kinases are a class of proteins that play essential roles in cell signaling, differentiation, and proliferation. They are frequently mutated in cancer and are the second-largest therapy target of specific inhibitors in clinical research. The activation status of mutated kinases in cancer can profoundly impact phenotypic outcomes not limited to tumor progression and drug sensitivity. To quantify these phenotypic outcomes through mutated kinase activities. To better understand and more importantly predict oncogenic activation of kinases at the molecular level, the role of mutations in intrinsic kinase activity needs to be quantified. We report enhanced sampling simulations, free energy calculations and several considerations form statistical mechanics to describe the various mechanisms of structural stabilization of kinases and their mutated systems and provided a detailed view of the underlying structure-activity relationship. Our results are validated against biochemical assays of kinase activation, cellular assays of transformation, and experiments probing structure and dynamics involving X-ray crystallography and hydrogen-deuterium exchange mass spectroscopy. We also describe a machine learning framework that integrates all of our results intro a predictive algorithm and show that it out performs the leading evolutionary and bioinformatics based algorithms.