Together, breast, colorectal, lung, ovarian and prostate cancer account for more than 850,000 cancer diagnoses and 290,000 deaths annually in the US. Family- and population-based studies have shown that these cancers have a heritable component, and multi-cancer susceptibility regions identified by genome-wide association studies (GWAS), suggests that this heritability is partly shared across cancers. Furthermore, GWAS have also revealed distinct susceptibility regions for specific tumor subtypes (e.g. estrogen receptor (ER)+ vs. ER- breast cancer), offering some insights to ethnic disparities in cancer incidence by tumor aggressiveness and mortality. This proposal will capitalize on the OncoArray data to study the genetic contribution to cancer risk and prognosis. The GAME-ON consortium recently launched the OncoArray initiative with the goal of identifying novel susceptibility loci for breast, colorectal, lung, ovarian and prostate cancer. In total, GWASdata on more than 350,000 individuals across these five GAME-ON cancers is being generated, creating unprecedented opportunities to jointly study multiple cancers in a homogenously derived dataset. We will quantify and functionally characterize the shared genetic contribution to GAME-ON cancers and use this information to fine-map multi-cancer GWAS regions. To accomplish our goals, we will leverage publically available databases (e.g. ENCODE) to functionally partition the shared genetic component, allowing us to assess the relative importance of functional categories (e.g. exonic, regulatory) for cancer development. For breast and prostate cancer, two cancers with ethnic disparities in cancer incidence and mortality, we will assess the genetic correlation between populations of European and African-American ancestry. Functionally characterizing the shared heritability between ethnicities will highlight biological mechanisms and provide insights into differences in disease aggressiveness between ethnicities. We will fine-map multi-cancer GWAS regions and estimate variant-specific posterior probabilities of causality. Our fine-mapping approach is unique in that it jointly models functiona annotations and the genotype-phenotype association while allowing for multiple causal SNPs within a region, greatly improving statistical power if more than one causal variant exist. Finally we will develop statistical methods to quantify the genetic contribution to variation in survival times and apply our methods on a prostate cancer dataset of 60,000 cases (9,000 deaths). Multi-generational studies show that a cancer prognosis is often "inherited" from parent to offspring suggesting a genetic component but lack of adequate statistical methods and empirical datasets have precluded formal assessment of such. This project capitalizes on the large-scale OncoArray initiative, giving us adequate statistical power to obtain precise heritability estimates Our results will elucidate cancer biology, identify novel connections between cancers and cancer subtypes and have long-standing impact on focus and design of futures studies aiming at understanding the mechanisms causing cancer.