Fig. 1: Accuracy of S-LDXR in null and causal simulations.

We report estimates of the enrichment/depletion of squared trans-ethnic genetic correlation (λ2(C)) in both null and causal simulations, for (a) quintiles of 8 continuous-valued annotations and (b) 28 main binary annotations (sorted by proportion of SNPs, displayed in parentheses). Results are averaged across 1000 simulations. Error bars denote ± 1.96 × standard error of the mean. Numerical results are reported in Supplementary Data 4 and 7.