Because multiple SNPs other than the CNPs at the CYP2A6 locus also showed significant association with CPD (Figure S4), we performed fine-scale imputation analysis for our GWAS set using the 89 Japanese subjects of the 1000 Genomes Project as a reference (Phase I 2011-11-23). We selected SNPs within a 1 cM region around the CNP (rs8102683), which encompasses a series of CYP2 family genes (Figure S4). Here, the common deletion at the CYP2A6 locus was addressed properly by the imputation study; we assumed that each SNP within the commonly deleted region consisted of two normal SNP alleles along with the common deletion allele. Therefore, the imputation treated the SNPs as tri-allelic markers (see Materials and Methods for details). As a consequence, none of the imputed SNPs showed greater significance than the CNP in the GWAS (Figure 2a), suggesting there are no causal SNPs in absolute LD () with the CNP.
