We identified 936 significant gene–phenotype relationships for binary traits and 767 for quantitative traits (Fig. 2a, Extended Data Fig. 3, Supplementary Table 8). These associations were enriched for FDA-approved drugs (binary odds ratio (OR): 7.38 (95% CI: 3.71–13.59), P = 1.46 × 10−7; quantitative OR: 3.71 (95% CI: 2.23–5.74), P = 7.04 × 10−9) (Fig. 2b, Extended Data Fig. 4; Methods) and spanned most disease areas and disease-relevant biomarkers (Fig. 2c, d). Many signals were of large effect, with a median OR of 12.4 for binary traits and a median absolute beta of 0.35 for quantitative traits. We also detected several significant genes with putatively protective PTVs, including APOB and PCSK9 (Supplementary Table 9). The median genomic inflation factor (λ) was 1.002 for binary traits (range: 0.71–1.35) and 1.010 for quantitative traits (range: 0.88–1.37) (Extended Data Fig. 5a). Only 0.76% of the associations from the 191,037 non-recessive collapsing analyses were outside the 0.9–1.1 λ range. Our tests were thus highly robust to systematic bias and other sources of inflation. Collectively, these findings provide biological insight into common diseases and substrates for future therapeutic development opportunities.
