For primary analyses, genome-wide significance was declared at P < 5 × 10−8. To account for multiple comparisons in analyses stratified by sex, genome-wide significance was also considered at P < 1.67 × 10−8. For genome-wide significant hits, Forest plots and PM-Plots were generated using the programs METASOFT with default settings and M-values were generated using the MCMC option13,78. For a given study and SNP, the M-value is the posterior probability that there is a SNP effect in that study. Studies with values <0.1 are predicted to have no effect, values ≥0.1 and ≤0.9 are ambiguous, and values >0.9 are predicted to have an effect. In PM-plots, M-values are plotted against -log10 P-values. Regional association plots were generated using LocusZoom79 with 400KB windows around the index variant and compared to the corresponding windows in the other ancestry groups, including the 1000 Genomes Nov. 2014 reference populations EUR, AFR and AMR, respectively. To test for sex-specific effects, a z-test was performed on the difference of the effect estimates from male and female sex-stratified analyses.
