The most compelling evidence for lack of bias is that our genome-wide significant finding was replicated in the independent GAIN cohort, which had data available on lifetime drug abuse histories for all of the defined cases and controls. Our discovery cohort also found nominal evidence of replication for a SNP previously reported to be associated with DSM-IV marijuana in a cohort using DSM diagnoses for identifying both cases and controls, further suggesting that the use of population controls functions acceptably for the study of drug abuse phenotypes.
