We enrolled 11,696 Japanese subjects in the GWAS for CPD (Table S1) with the support of the BioBank Japan Project [7]. Stringent quality control criteria for both CNPs and SNPs, including principal component analysis (PCA), were applied as described previously [8], [9]. To extend the genomic coverage, genome-wide imputation was performed for SNPs using data from HapMap samples (JPT + CHB; Phase II). Consequently, the genotype data for 4,256 autosomal CNPs and 2,312,503 autosomal SNPs with minor allele frequencies (MAF) 0.01 were obtained (see Materials and Methods for details). Each CNP or SNP was then evaluated for association with CPD using a linear regression model that accounted for the additive effects of copy number dosage or allele dosage on CPD with other covariates. Although no significant population stratification was suggested by the data from our study population (Figure S1), we also used the first two eigenvectors within the East Asian population (Figure S2) as covariates. The Quantile-Quantile plot of the -values exhibited an inflation factor (; [10]) of 1.01 for the genome-wide SNPs (Figure S3a), which suggests that there was
