Project Summary Cancer genomics has transformed the study and treatment of cancer. Nevertheless, the interpretation of genetic alterations observed in tumor samples remains a major challenge in both basic science and clinical settings. The functional roles of the vast majority of somatic alterations observed within cancer cells are unknown: it is typically not known if a specific mutation plays a role in cancer growth or anti-cancer immune response, or what impact (if any) an observed alteration has on cellular pathways. These gaps in knowledge are major impediments towards personalized cancer treatments, from uncovering actionable mutations that can be targeted by drugs to designing personalized cancer neoantigen vaccines. Further, for diverse population groups to reap the benefits of cancer genomics, it is critical to develop computational methods that explicitly consider genome variation and that are designed to identify and potentially mitigate inequities. Our long-term goal is to develop computational approaches that work well across diverse populations in identifying medically relevant alterations within cancer genomes. We have developed a powerful suite of methods to discover cancer driver genes based upon integrating observed somatic mutations with knowledge about cellular networks and a multidimensional view of protein functionality at the level of individual sites. We will build upon on our prior work by devising machine learning approaches to predict which specific somatic mutations within genes drive cancer, along with software to visualize and interpret mutations with respect to newly derived protein functional features. We will introduce integrative formulations to uncover the impact of somatic alterations on downstream genes and pathways. We will devise approaches to uncover immune- relevant somatic mutations based on predicting interactions between major histocompatibility complex proteins and peptides mutated in cancer. We will design our approaches to be effective as well as equitable across the diversity of human populations. We will release open-source implementations of the developed computational methods. This project will advance the state-of-the-art in computational methods for interpreting cancer genomes, deepen our understanding of cancer biology, and move the field further towards unlocking the promise of personalized cancer treatments in an equitable manner.