are significant when we considered the entire genome, but, when we focused on specific gene regions, they are not statistically significant. The result observed, considering genomic differences, is in accordance with recent evidence that has indicated differences in the stratification of functional RVs among human populations [40,41]. As is the case for common variants, the differences in RVs between African and non-African populations may be attributed to human demographic history. However, the specific-gene analysis suggests that, in drug dependence GWAS, confounding effect due to rare variation is not linked to the ancestry of the investigated populations. Nevertheless, we cannot exclude that RVs could have a population-specific effect on an SD-phenotype, since their effect could depend on a permissive epistatic genetic background present in one population group but not another. Moreover, the high rates of functional RVs in both GWAS-relevant genes and their interactive partners also indicate that rare variation may strongly confound outcomes of drug dependence GWAS, as postulated for other complex phenotypes related to drug response [42,43].
