Since many of these tests are correlated (many SNPs are in substantial LD and there is overlap of markers in multi-marker analyses), Bonferroni correction seems excessive in this case. Based on this rationale, we decide to use permutation tests to obtain empirical p values for the multi-marker haplotype tests. With 10,000 permutations, combinations 5-6-7, 5-6-7-8 and 1-3-5-6-7 remain significant for SI (p = 0.0090, 0.0114 and 0.0259 respectively). Based on these results, we may reasonably conclude that OPRM1 is associated with smoking initiation. Whether OPRM11 is associated with nicotine dependence is less clear. Further studies with independent samples are necessary to resolve these issues.
