Project Summary Many phenotypes, as well as the risk of developing many diseases, are genetically complex, and involve contributions from both genetic and non-genetic factors. Work in human genetics over the past two decades have shown that this variation is the result of contributions from a very large number of sites, on the order of thousands or tens of thousands. This presents challenges for both the measurement and interpretation of genetic association studies, as real genetic effects can be difficult to distinguish from the effects of confounding biases. On the other hand, biobank scale resources represent a tremendous opportunity to learn about both the biology of complex traits, and the evolutionary forces that have shaped modern patterns of variation. My group will develop statistical methods to overcome several current challenges in the study of genetically complex traits by apply tools from population, quantitative and statistical genetics. First, we will develop tools to diagnose and correct for ancestry stratification biases in polygenic scores. Even subtle stratification biases compound across loci to cause problems with polygenic predictions, so methods of carefully accounting for these biases are needed. Second, we will study the role of mutational pressure in maintaining complex disease and shaping its genetic architecture. The increasingly availability of exome and genome-wide sequencing association datasets make estimating the strength of mutational pressure toward increased disease risk increasingly feasible. New theoretical development will be needed to make and interpret these measurements. Third, we will develop models to study how mutation and selection jointly shape the distribution of heritability for complex traits across genomic regions with different functions. Current methods confound these two effects, so there is an opportunity for principled population genetic modeling to provide clarity on the biology of complex traits. Finally, we will develop improved methods for coalescent inference in population genetics. Recent breakthroughs in coalescent inference have begun to reshape our ability to learn about evolutionary events from genome sequencing data. However, these methods exhibit clear accuracy-scalability tradeoffs, suggesting that a thoughtful approach to inference is needed if the benefits of these methods are to be fully realized. My group will develop methods for accurately estimating coalescent times from sequencing data.