These 25 CNVRs were followed up by investigating an independent case–control GWAS sample: 20 of the 25 CNVRs also indicated evidence for a directionally consistent association with obesity (one-sided P ≤ 0.025; 5 of the 20 CNVRs with a Bonferroni-corrected CNVR-adjusted one-sided P ≤ 0.025; Table 2). Performing 10 000 permutations of the case–control status, we observed only two permutation samples with 20 or more of such directionally consistent associations with obesity (empirical P-value 2 × 10−04). For 3 of the 20 CNVRs [CNVR 10q11.22 (1), CNVR 16p13.11 and CNVR 17q21.31], association signals for obesity were found at exactly the same CNV marker in both GWAS discovery samples. In the remaining 17 CNVRs, at least one CNV marker showed evidence for an association with obesity in the family-based GWAS sample, while at least one other correlated CNV marker in the case–control GWAS sample similarly showed evidence for an obesity association signal. The maximal pairwise correlation coefficient between each of these two CNV markers associated with obesity ranged from 0.24 to 0.91 as estimated in the parents, whereas the range of
