CNVRs and the surrounding ± 500 kb DNA segments with evidence for an association to obesity as based on the CNV-FBAT (family-based GWAS sample; at least one marker with a two-sided asymptotic P-value ≤ 0.05) and with additional evidence for association as based on the logistic regression (case-control GWAS sample; at least one marker with a directionally consistent two-sided asymptotic P-value ≤ 0.05), were analysed with the aim to identify SNPs that are in linkage disequilibrium with the involved CNVs (see 26). It was reported that CNVs that were well genotyped were frequently tagged by SNPs (26). Again circumventing the calling, we determined Pearson correlations of intensity values for CNVs and SNPs in the set of all 424 parental pairs of the GWAS trios. For a CNVR-tagging SNP we required that the squared Pearson correlation coefficient (r2) of intensities had to be >0.6 for at least three consecutive CNV markers. CNVR-tagging SNPs were detected for two of the 20 CNVRs assessed. These two SNPs (rs9804659, rs2815752) were followed-up for replication association analyses in the sample of 365 independent case-parents obesity trios.
