We applied genomic control correction for low-frequency variants (MAF < 1%) in both multi-ancestry and ancestry-stratified meta-analyses. Genomic control correction for common variants was not applied given that elevation of genomic control values is expected with high polygenicity (that is, it assumes sparsity) and very large sample sizes48; such a correction may be overly conservative. To evaluate this decision, we estimated the replicability of associated loci using a trans-ancestry extension of an existing method6. This method, ‘RATES’, incorporates cohort-level summary statistics (single-nucleotide polymorphism (SNP) effect sizes and their corresponding standard errors), along with allele frequency-based MDS components per study to assign a posterior probability that each sentinel variant effect would replicate in a sufficiently powered study. To further evaluate robustness of our results, we estimated LD score regression (LDSC) intercepts and attenuation ratios to account for bias in the intercept test when sample sizes become extreme, as in the present case. Results were within expected limits and consistent with a limited effect of population stratification on the meta-analysis results44 (Supplementary Table 8). Then, we compared the sign of SNP effect
