This study has limitations. The sample size (number of Ca-HL cases) is small in the discovery GWAS, indicating that the power of our GWAS is likely to be low except for risk loci of large effect. Genes close to other prominent SNP association signals (P < 5 × 10−7) in the meta-analysis of Yale-Penn samples may warrant further study, including APBA2, EFCAB3, MTRR, PRPRN2, SPATA6, and FLJ12825 (SNPs mapped to these genes are included in Supplementary Table S5). A continuous measure (e.g., frequency of having experienced hearing, seeing or smelling things that were not really there after marijuana use) would provide more statistical power and should be considered for future surveys eliciting information on the trait.
