response was rated on a visual 100 mm scale that was then converted to a rating from 1-10. Blood cortisol levels were collected at five time points described in Figure 3. Log-transformed cortisol levels were analyzed in mixed models that included the interaction of FKBP5 genotype with time point rather than experimental dose. MNWS, PANAS and BQSU data were analyzed at the pre- and post-nicotine conditions separately with age, sex, and race as covariates. DSM-IV nicotine withdrawal symptoms were summed, log-transformed and analyzed in regression models that used age, sex, anxiety or depression diagnosis (coded dichotomously), the first three PCs (determined for the EA and AA populations separately), and rs3800373 (coded as minor allele carrier versus TT) as predictors. The results for EAs and AAs were combined by meta-analysis. Analyses and quality control procedures were performed using PLINK40 and JMP® Pro 10.0.
