of cases + 1/No. of controls).40,41 We used effective sample sizes due to the small percentages of cases in these cohorts. For the UKBB dataset, we used an AUDIT score of 12 or greater (range, 0-40) to determine the number of cases for the purpose of calculating the effective sample size as suggested by the authors. Only SNVs with the same directions of effects in all 3 GWASs as determined with METAL software40 were kept.
