Stabilising selection enriches the tails of complex traits with rare alleles of large effect

Establishing the relative contribution of common and rare variants to complex trait heritability is a key goal of biomedical research. Recent statistical genetics inference suggests that common variants explain most complex trait heritability, but little is known about how genetic architecture varies across the trait continuum. If rare variants make a small contribution to heritability but have their effects concentrated in the tails of complex traits, where disease typically manifests, then they may have a greater clinical impact than previously inferred. Here, we perform simulations using the forward-in-time simulator SLiM to generate realistic population genetic and complex trait data, in which traits evolve under neutrality or stabilising selection. Recent studies suggest that stabilising selection is the dominant force shaping the genetic architecture of complex traits, consistent with our simulations in that data simulated under stabilising selection here more closely resembles real data. Moreover, we observe a shift of rare, large-effect alleles towards the tails of the complex trait distribution under stabilising selection. In our simulations, individuals in the tails of complex traits are, depending on the strength of selection, 10-20x more likely to harbour singleton or extremely rare alleles of large effect under stabilising selection than neutrality. Such an enrichment of rare, large-effect alleles in the tails of real complex traits subject to stabilising selection could have important implications for the design of studies to detect rare variants, as well as for the prediction and prevention of complex disease.