In the UKB, quality control was performed on the alcohol consumption phenotype to remove extreme values. Weekly intake values of >5 s.d. from the mean were set to missing. For males this was for values >102 units per week and for females for values >89 units per week. Alcohol consumption was then log (units+1) transformed and, as the mean alcohol intake in males was significantly higher than in females, we regressed age and weight in kg onto weekly units of alcohol consumed in males and females separately. We then took the residuals from these regressions and pooled the male and female residuals together to create the alcohol consumption phenotype. A male-only and female-only GWAS was also performed.
