The impact of demography and selection on the genetic architecture of human traits

About This Grant

Project summary The human population has complex genetic structure because of past and ongoing patterns of migration and admixture, because of nonrandom mating based on heritable traits, and because of selection on disease and non-disease traits. These processes have profound consequences not only for our understanding of human genetic variation, but also for our understanding of the genetic architecture of human traits, including those relevant to human health and flourishing. The genetic architecture of a trait can be summarized as: (i) the loci that underlie variation in the trait; (ii) the effects and frequencies of the alleles at these loci; and (iii) the correlations between alleles across these loci. My work seeks to understand how these features of genetic architecture interact with one another, and how they—and our ability to learn about them from genomic data— are impacted by the processes that determine the genetic structure of human populations. In the first part of my proposed research, I will develop a novel genomic test for stabilizing selection (selection against extreme values of traits) and assortative mating, based on the different effects of recombination on the allelic correlations across loci—long-range linkage disequilibria (LD)—that stabilizing selection and assortative mating generate. I will use this LD–recombination signature to jointly estimate the strengths of stabilizing selection and assortative mating in genome-wide association study (GWAS) data for human traits. I will furthermore characterize how the long- range LD generated by stabilizing selection slows down the allele-frequency dynamics expected under stabilizing selection. I will interrogate the consequences of this slowdown for the equilibrium distribution of allele frequencies and effect sizes, and for the portability of genomic studies across populations. Second, I will seek to understand how demographic processes and selection impact the GWASs that we often use to learn about genetic architecture. Traditional methods to control for the biases that population structure can generate in GWASs have recently been shown to be inadequate in important test cases. I will use coalescent-based approaches to reframe these methods in terms of their ability to capture long-range LD structure in the genome. I will then use this theoretical framing to understand the shortcomings of these methods by studying the consequences of various demographic and selective forces on long-range LD. Then, I will develop and analyze a GWAS design that uses chromosome-specific polygenic scores to control for non-random mating and selection—which traditional methods struggle to control for—in population-based GWAS. Third, I will show that stabilizing selection on complex traits generates selection against the minor-parent ancestry in admixed populations. Using comprehensive simulations of demographic and selective scenarios relevant to Neanderthal–human introgression, I will explore whether this novel mechanism of selection against introgression can explain the extent and pattern of Neanderthal ancestry in the human genome. Finally, I will characterize the genomic signals of introgression unique to this mechanism of selection, and test for them in biobank-scale data.