We analysed 1,419 quantitative traits in a linear regression model including individuals of all major ancestral groups, including Europeans (Supplementary Table 1). This model included categorical ancestral groups, the top five ancestry principal components, age and sex as covariates (Methods). We identified 59 significant gene–quantitative trait associations that were originally not significant in the European analyses (Fig. 3b, Extended Data Fig. 6b). These included associations between rare variants in OCA2 and a younger age of wearing glasses (P = 4.7 × 10−10; β: −0.45 (95% CI: −0.60 to −0.31)), ASGR1 and reduced low-density lipoprotein cholesterol (P = 1.7 × 10−9; β: −0.26 (95% CI: −0.34 to −0.17)), and others (Supplementary Table 17). In addition, 46 unique associations between genes and quantitative traits, originally significant in the European analyses, were not significant in the combined analysis.
