11 0.85–1.46 rs7338244 37065052 Intron 2 C/G 0.306 1.000 0.003* 1.34 1.11–1.63 0.015 1.33 1.08–1.71 >0.1 1.27 0.95–1.66 Two SNPs remained statistically significant after correction for multiple testing using FDR method. OR >1, the reference (minor) allele is associated with the higher risk of low BMD B36 Genomic position, MAF minor allele frequency, HWE Hardy–Weinberg equilibrium,

selleck kinase inhibitor LS lumbar spine, FN femoral neck *P FDR < 0.05 Association between POSTN and BMD variation in tSNP-based analysis Table 2 lists the single-marker association results with BMD variation in the extreme cohort. Two tSNPs (4-Hydroxytamoxifen chemical structure rs7322993 and rs7338244) in the POSTN gene showed significant associations with BMD variation after the correction of multiple testing (P FDR < 0.05, OR > 1). Both of their minor alleles were related to the higher risk of low BMD. SNP rs7322993 had the strongest association (P = 0.001), and the P values were 0.006 and 0.029 for BMD at LS and FN, respectively, in site-specific analyses. We examined the association between common haplotypes of POSTN and BMD variation. The LD structure of POSTN is illustrated in Figure S1 (ESM EPZ5676 order 1). The haplotype in POSTN was associated

with BMD variation in the global test (P = 0.038) (Table S2, ESM 1). Table S2 (ESM 1) also shows the specific effects of each haplotype. Seven haplotypes were identified, each with a frequency >3%. The haplotype CGTTGAAG, including both the low BMD-related alleles of rs7322993 and rs7338244, was most strongly associated with BMD variation (P = 0.025) and a higher risk of low BMD was expected. The haplotype data corresponded to the single marker analysis, but did not add further information to outcome. Validation of rs9547970 with imputation-based association testing We used the imputation method to study SNPs that were not genotyped in our study, but available from the HapMap database. This enabled a more comprehensive fine map of polymorphisms along POSTN and a better understanding of this association. We estimated

the strength of evidence for a phenotypic association with typed Cobimetinib price and untyped SNPs located between ∼5 kb upstream and ∼5 kb downstream of POSTN. Sixty-two SNPs from the Asian population data of HapMap phase II were extracted for imputation. In addition to the genotyped SNPs, 54 imputed SNPs (MAF ≥ 0.01 and the imputation quality r 2  ≥ 0.3) were tested for associations with BMD variation. The strongest evidence for an association with BMD variation in the imputed data is undoubtedly for the untyped SNP rs9547970 (P FDR < 0.05), which is located at −2,327 bp upstream of POSTN (Fig. 1). An additional 27 SNPs displayed convincing evidence of association (P < 0.005) and were in high LD (r 2 > 0.5) with rs9547970 based on the HapMap Asian population data. The most significant untyped-SNP rs9547970 had a high imputation quality (r 2 = 0.9983). Subsequently, it was also directly genotyped in the 1,572 extreme samples to verify its association with BMD variation.