we found an excess of the same alleles among cases as observed in a prior report (Levran et al. 2008) in a predominately Caucasian sample with some Middle Eastern contribution , we consider this possibility unlikely. We also reran analyses increasing the number of principal components to control for admixture to ten (from four), with little effect on results. Despite its widespread use as a method to control for multiple testing, spectral decomposition may be viewed by some as inadequately conservative. It is thus important to note that when a more stringent threshold such as a Bonferroni correction (i.e., 0.05/136= 3.68 × 10−4) is applied, our two most highly associated SNPs remain significant. Despite the considerably larger size of our sample (more than three-fold larger than most prior association studies of heroin dependence), it is possible that we may have failed to detect significant associations because of limited power (i.e. type II error). Similarly, the relatively smaller size of the neighborhood control sample and any sharing of risk with related phenotypes (e.g., other substance dependence or externalizing disorders) more prevalent in this group (than the non-dependent controls) could be contributing to the lack of significant differences found in this comparison.
