Current methods development is done within the context of the IMAGE Program Project: The overall goal of this research program is the integration of different types of information to build comprehensive statistical models for cancer causes and prognosis. Motivated by various studies of colorectal cancer, we are developing novel statistical methods that use prior biological knowledge to inform integrative genomics analyses, that use phylogenetic information to infer gene function, that incorporate high-dimensional data on the internal environment (the microbiome and the exposome), and that model tumor evolution from data on somatic changes within and between tumors. Beyond colon cancer, our methods are broadly applicable to other cancer types and many other chronic diseases.