In this presentation we will describe a number of different oncology programs focusing on how population cohorts, rather than clinical cohorts, enable studies of early detection and evolution of cancer. Early cancer detection is critical to improving health outcomes and cancer survival. Our team leads a number of programs that have generated genomic and machine learning technologies and resources that can be adapted to improve early cancer genomic studies and biomarker development for both blood, as well as solid tissue tumors. Recent studies have linked age-related mutation accumulation in blood to a number of diseases. There is a clear need to understand the biology and develop biomarkers and tools that distinguish benign accumulation from cancer. Recent evidence from our team and others shows that in a subset of individuals, age related clonal hematopoiesis (ARCH) leads to the generation of pre-leukemic hematopoietic stem cells (HSCs) and eventual leukemia development using population cohorts. Finally, we show how utilizing cell-free DNA as a screening tool for early cancer detection requires profiling of blood plasma samples collected from asymptomatic individuals prior to the diagnosis of cancers to enable assessment of the earliest detectability and predictive performance of potential biomarkers. We demonstrate that cfDNA methylation markers are indicative of breast cancers are detectable up to 7 years prior to a stage I diagnosis.