In addition to sources of genomic variation (e.g., allele frequencies, LD), distinctions in findings across the ancestral groups are possibly attributable to the pattern of comorbidities in these groups, which may be genetic and environmental in nature. Notably, a fair proportion of AA qualified for a diagnosis of ANYDEP due to cannabis or cocaine dependence, whereas the preponderance of EAs primarily endorsed alcohol dependence. In addition, cannabis and cocaine dependence diagnoses in AA were relatively more severe (i.e., more criteria were endorsed). Furthermore, drug_noalc (h2=0.63) was more heritable than ANYDEP (h2=0.37) and alcohol dependence (h2=0.27) itself in the AA but not the EA families. Thus, despite the smaller sample size, the AA subsample may have been better powered to identify loci more closely related to drug dependence. These patterns of individual and comorbid drug use disorders are also quite consistent with the broader epidemiological literature3,21. For instance, AA are more likely to initiate use of cannabis prior to alcohol and are more likely to escalate to problem use60,61. Similarly, AA are at nearly 3.5 increased odds of transitioning from cocaine
