Polygenic signals of sexually antagonistic selection in contemporary human genomes

Mutations that increase fitness in one sex may decrease fitness in the other. Such “sexually antagonistic” (SA) genetic variants can constrain adaptation and increase variability for fitness components (e.g., survival, fertility, and disease susceptibility). However, detecting SA selection in genomes is immensely challenging, as it requires prohibitively large datasets that combine genomic sequences with individual fitness measurements. Here, we use genotypic and reproductive success data from ∼250,000 UK Biobank individuals to comprehensively assess the extent of SA genetic variation in humans. We first develop new theoretical models for signals of SA selection spanning a full generational life cycle—including SA polymorphisms affecting survival, reproductive success and overall fitness. Comparing our models with UK Biobank data, we uncover multiple empirical signals of polygenic SA selection, including sex-differential effects of genetic variants on each fitness component, and positive correlations between sex-differential effects and minor allele frequencies. We show that these signals cannot be explained by simple models of sex differences in purifying selection, or by potential confounders such as population structure and sequence mapping errors. We further show that candidate SA sites disproportionately affect functional genomic regions, including polymorphisms associated with quantitative traits and disease. Finally, we examine historical evolutionary processes affecting candidate SA sites, which are consistent with the drift-dominated dynamics predicted by previous theory. Overall, our results support SA genomic variation in humans and highlight its broader functional and evolutionary consequences.