For our PheWAS, we investigated multiple alleles mapped to the CYP2A6 locus based on previous evidence of nicotine-related functional effects. These variants were selected from the significant results of a recent genome-wide association study (GWAS) of NMR8 and from known functional CYP2A6 alleles from the Human Cytochrome P450 (CYP) Allele Nomenclature (available at http://www.cypalleles.ki.se/cyp2a6.htm). For the WHI cohort, we considered the WHI SHARe dataset (phs000386.v5.p3) because it includes the genome-wide data needed to impute CYP2A6 alleles. A detailed description of the procedures used for genotyping, genotype quality control, principal component analysis, and imputation of the WHI dataset is found in our previous PheWAS12. Briefly, we conducted a principal component analysis using Plink 1.922 and genome-wide datasets pruned for linkage disequilibrium (r2 > 80%). Genotype imputation was performed using SHAPEIT23 for pre-phasing, IMPUTE224 for imputation, and the 1000 Genomes Project Phase 325 as the reference panel. After imputation, we obtained good-quality genotype dosage information for nine CYP2A6 variants (info score > 0.5; Supplemental Table 2), including three independent genome-wide significant SNPs identified by the NMR GWAS8 and six functional CYP2A6 alleles
