When focusing on CNVRs, comparability of association findings across samples needs to be addressed as different individual CNV markers might contribute to the association in different samples. We determined pairwise Pearson correlations between pairs of selected CNV marker intensity values in each of the two discovery GWAS samples. The selection was based on the presence of some evidence for an association with obesity for the CNVR (family-based GWAS: two-sided asymptotic P-value in CNV-FBAT ≤ 0.05; case–control GWAS: directionally consistent two-sided asymptotic P-value in logistic regression ≤ 0.05). We report the maximal Pearson correlation coefficient (r) computed from the set of selected CNV markers (Table 2, 10th column). For comparability, we also report the Pearson correlation coefficient (r) for those two CNV markers with minimal P-values in the family-based and in the case–control discovery GWAS samples, respectively (Table 2, 9th column).
