PRS were split into quartiles and odds ratios calculated for case status for each quartile of risk compared to quartile 1 (lowest). A. Alcohol dependence in ALSPAC was coded by DSM-IV diagnosis (484 cases, 3,837 controls). B. Alcohol dependence was coded via DSM-IV diagnosis in the COGA sample (2,318 cases, 4,532 controls). C. A sample of European ancestry, unrelated British individuals who had ever drank, had ICD9 and ICD-10 codes available, and were not included in the discovery GWAS were used (4,141 cases, 241,806 controls).