Single Nucleotide Polymorphism Spectral Decomposition (Nyholt 2004; Li & Ji 2005) was first used to calculate the appropriate correction for multiple testing for analyses of data for the 136 opioid receptor SNPs that remained after data cleaning. As the respective linkage disequilibrium (LD) plots demonstrate (see Figure 1, Supplementary Figures 1 and 2), we genotyped a number of SNPs in high LD for these genes. Based on the calculated effective number of independent loci markers (57.69), the significance threshold necessary to keep the type I error rate at 5% was determined to be 8.89 × 10−4. Association analyses were then performed using PLINK (Purcell et al. 2007). Logistic regression, including the smartpca-derived PCs in the model as independent variables to control for admixture, was used to examine the association between the log-additive effects of minor allele dosage and status (case versus non-dependent control). Although PC covariates were not needed for the secondary comparisons of cases versus neighborhood controls (see PCA results below), logistic regression was again used to obtain results comparable to the above.
