In this study, we provided a catalog of amino acids involved in known somatic mutations within pocket regions and across cancer types. Our systematic analyses revealed that two amino acids, Arg and Glu, were most frequently mutated within pocket regions across multiple cancer types. Specifically, Arg mutations were attributed to the anti viral immunity and cell cycles of APOBEC3G, which is consistent with previous mutational signature analysis study. Several recent studies, such as SpacePAC, iPAC, and GraphPAC, identified mutational clusters in cancer by integrating somatic mutation data and protein structure information. In comparison with these studies, our protein pocket based approach provides an alterna tive to identifying actionable mutations in the pocket re gions that are attributed to tumorigenesis, and further, to anticancer drug responses.

In summary, our protein pocket based integrative analysis provides important in sights into the functional consequences of somatic mu tations in cancer. There are several limitations in the current work. First, the somatic mutation profiles from both the COSMIC and TCGA are mixed with driver and passenger muta tions. Second, our approach requires protein 3D structural information to accurately detect protein pocket regions. The current protein pocket information is far from complete and may be inaccurate, due to the feasibility of protein structures. Although about 100,000 protein and nucleic acid structures have been curated in the PDB database, the human protein 3D structure information is still far from being sufficient.

In the future, we propose to improve our work in the two following ways use the experimentally validated driver mutations and passenger mutations from Vanderbilts MyCancerGenome database to investigate the functional roles of driver mutations versus passenger mutations in protein pocket regions and non protein pocket regions, and integrate homology modeling protein pocket information from other organ isms, as well as protein interface information in protein interaction network, large scale atomic resolution protein network, and protein post translational sites , to deeply ex plore the functional consequences of somatic mutations altered protein function in cancer.

Despite its limit in the scope of the current investigation, the data allowed us to systematically explore the roles of somatic muta tions in protein function and drug binding/response through a protein pocket prioritization approach. As a proof of principle Dacomitinib study, we demonstrated that the pro tein structure based strategy is a promising approach to gain insight into the functional consequences of somatic mutations in cancer. Conclusion Detecting actionable mutations that drive tumorigenesis and alter anticancer drug responses is in high demand in molecular cancer research and cancer precision ther apy.