Amplification is the primary mode of gene-by-sex interaction in complex human traits
Authors: Carrie Zhu 1,2; Arbel Harpak 1,2
Affiliations: 1) Department of Integrative Biology, University of Texas at Austin, Austin, TX; 2) Department of Population Health, Dell Medical School, Austin, TX
Keywords: Complex traits
Sexual dimorphism is observed in many complex traits and diseases and is suspected to be in large part due to widespread gene-by-sex interactions (GxSex). To date, empirical evidence for GxSex in GWAS data—with attempts focused on identifying imperfect genetic correlations between the sexes— has been underwhelming. We hypothesized that GxSex may indeed be pervasive, but largely missed by current approaches, because it acts primarily through sex differences in the magnitude of many genetic effects (“amplification”), regulated by a shared cue such as a sex hormone, rather than differences in the identity of causal variants or the direction of their effect (“correlation”). To test this hypothesis, we inferred the genetic covariance structure between males and females across dozens of traits in the UK Biobank, and decoupled GxSex signals into correlation and amplification effects. Amplification was a much more pervasive mode of GxSex. For example, we estimate that 38% of causal variants have a greater effect on urate levels in females than males. In addition, we investigated whether testosterone levels, a continuous measure, may underlie the observed mediation by binary biological sex. For many traits, notably in body mass related traits, testosterone levels are strongly associated with the magnitude of genetic effects in both males and females—but the association is different in magnitude and sign between the sexes. Finally, we developed a novel test of sexually-antagonistic selection linking GxSex signals from GWAS and allele frequency differences between males and females. Our test can help identify specific complex traits that are under contemporary sexually-antagonistic selection. In summary, our results suggest that the systematic amplification of genetic effects is a primary mode of GxSex that is crucial for understanding sexual dimorphism and may be germane for phenotypic prediction.