Population genetics of human evolutionary history and natural selection

This proposed research addresses unresolved and fundamental questions in human evolutionary history and genome biology. It contains three related research themes: (1) uncovering key features of human population history and structure, both in the recent and ancient past, (2) determining factors influencing selection across the genome and the limits to cross-population phenotypic prediction due to population history and natural selection, and (3) modeling and predicting the genetic basis of complex traits under realistic models of human evolutionary history, population structure and admixture. This work combines ambitious but feasible methodological developments with new sequencing data from worldwide populations. Human evolutionary history is imprinted in the DNA of people living today. Learning this history involves the combined efforts of theoretical, statistical and empirical analyses. As the volume of genome sequencing data grows, we have the opportunity to infer new details from key periods in our species’ evolution and clarify how population biology and natural selection shape phenotypic and genetic variation. This requires developing population genetic theory and computational methods that can handle data from many populations and large sample sizes, and that simultaneously incorporate these many different factors that impact human genetics. Over the next five years, my group will develop theoretical and methodological innovations that enable new discoveries from human population genomic data, and we will uncover details of human history that resolve important questions about our species’ evolution. This work builds on my lab’s expertise in population genetic theory and simulation methods. Recent work shows that our newly developed approaches for modeling variation and linkage disequilibrium statistics are powerful to learn complex multi-population history. Combined with newly sequenced high-quality genomic data, we will further our understanding of early human evolution and reconstruct recent admixture and demographic histories. These models contribute to genomic resources that are essential to downstream studies to understand genetic components of human health and disease susceptibility. Our theoretical advances will enable new analyses of the combined effects of selection and recombination in multiple populations, which we will use to understand fundamental limits to the portability of phenotypic prediction across populations and in recently admixed groups. Finally, we will develop approaches to predict the architecture of polygenic traits in structured and admixed populations, settings that are most relevant for human genomic studies but current methods fail to address. The mathematical and statistical advances from this work will be developed into open source, maintained computational resources that will facilitate genomic discoveries in our own research and that of others in the years to come.