Within-family effect of ancestry on complex traits in a Mexican population

Human populations differ in disease prevalence and phenotypes, but the extent to which differences are caused by genetic factors is unknown for most complex traits. Comparing phenotypic means across populations is confounded by environmental differences and using polygenic predictors can lead to biased inference^1,2. Family-based analyses of people of genetically admixed ancestry enable estimation of ancestry effects unconfounded by ancestry-environment correlations. Here we leverage genetic data from admixed adults in the Mexico City Prospective Study^3,4 to estimate within-family ancestry effects⁵. We assessed genetic ancestry and 15 complex traits in 52,583 unrelated people and 39,714 relatives from 17,627 families. At the population level, relative to European ancestry, the effect of Indigenous American ancestry was -1.98 s.d. (P < 2 × 10^-16) for height and a natural log odds ratio of 1.73 (95% confidence interval, 1.54-1.92) for type 2 diabetes. Within families, the effect of Indigenous American ancestry was -1.51 s.d. (P = 10^-8) for height and natural log odds ratio of 5.13 (95% confidence interval, 2.48-7.78) for type 2 diabetes. These effects are supported by between-ancestry differences in trait-increasing allele counts and evidence of selection at trait-associated loci. We found no within-family ancestry effect on educational attainment or other traits despite significant associations at the population level, implying environmental causes or confounding. Overall, this study provides an experimental design to study between-ancestry genetic effects and identifies significant ancestry differences for height, type 2 diabetes and metabolic traits in a genetically diverse population from Mexico City.