A subset of individuals from the American Cancer Society (ACS) Cancer Prevention Study-II Nutrition cohort (N ~ 184,000) was used as a replication dataset. DNA was collected from 109,380 individuals from this cohort. Smoking behavior information was obtained from questionnaires administered in 1982, 1992, and 1997. Based on the responses to these questionnaires, a case/control phenotype (heavy smokers/light smokers) was developed as a proxy for nicotine dependence and non-dependence (Stevens et al. 2008). Inclusion in the ACS dataset used for our analysis required subjects to report smoking at least 100 cigarettes lifetime. Cases (heavy smokers) also needed to report in at least one interview smoking at least 30 cigarettes per day for at least 5 consecutive years. Controls (light smokers), in addition to smoking more than 100 cigarettes lifetime, smoked for at least 1 year, but endorsed the lightest smoking rate available on each of the three questionnaires (<5 cigarettes/day in 1982 and 1992, <10 cigarettes/day in 1997). After ascertainment for heavy and light smoking, 1,500 heavy smokers and 1,500 light smokers were selected for genotyping. Of these, 2,847 smokers
