We are aware that these results could be false positive ones, but in order to minimize the likelihood of it, we used the best available methodology and analysis. Firstly, the investigated sample is one of the largest samples of crack/cocaine patients reported in the literature to study genetic susceptibility for cocaine dependence to date. We also adjusted for confounding factors (age and sex) to confirm that the association was not due to bias. Population stratification was tested with the EIGENSTRAT software and no differences were observed between cases and controls which could hinder the association results. In contrast with other studies investigating the association of genetic markers and cocaine dependence [31], [32], subjects in the present study were stratified by their preferred route of drug administration. This may decrease the sample's clinical heterogeneity, which frequently reduces the power of genetic association studies. Another confounding factor in previous studies was a concomitant dependence on other drugs, mainly alcohol and heroin. In the present study, individuals with alcohol dependence were excluded, and heroin dependence is very rare in Brazil (less than 1%
