After the analysis of each dataset, we carried out an inverse-variance meta-analysis on the results using METAL26, applying a genomic control to adjust for inflation in each dataset. Of the 751 suggestive SNPs (P < 1×10−5), we found these SNPs to show lower standard errors and confidence intervals with the increasing number of cohorts showing consistent directionalities of effect. Particularly, the average standard error for SNPs showing 1 to 7 consistent directionalities ranges from 0.171, 0.109, 0.0744, 0.0346, 0.0234, 0.0173 to 0.01795 (Supplementary Fig. 1b). Thus, we limited our final analysis to SNPs that showed consistent directionalities of effect in at least 6 out of the 7 datasets included in the meta-analysis. The association graphs of results from loci of interest were plotted using LocusZoom75.
