For quality control, we removed SNPs with a missing rate > 0.01. We tested for consistency with Hardy-Weinberg Equilibrium expectations and excluded SNPs with P-value < 0.0001. SNPs with minor allele frequency (MAF) < 5% were also removed to focus on the analysis of common variants. Genetic relationships were examined in the family-based sample by calculating pairwise identity by descent (IBD) proportion estimates using PLINK [21]. Thirty-six subjects with missing phenotypes were excluded in our analysis. In addition, 407 subjects with alcohol abuse were removed because their affection status was uncertain. After all data cleaning and quality control (QC) was completed, there were 2875 individuals and 769,498 SNPs for analysis. Finally, our AD sample consists of 1156 controls (405 males and 706 females) and 1719 cases (1004 males and 715 females). Among the 1156 AD controls, there are 178 and 612 cases of opioid dependence and cocaine dependence, respectively.
