We followed CNVR 11q11 up in the replication sample of 365 independent obesity trios due to the availability of a CNV-tagging SNP (rs9804659) and observed a directionally consistent trend for an association to obesity with a one-sided P-value of 0.076 [odds ratio for the obesity effect allele (OR) = 1.188; 95% confidence interval (CI) = 0.939–1.502]. The maximal squared Pearson correlation (r2) between rs9804659 and any of the CNV markers in CNVR 11q11 was 0.65, while r2 was 0.53 between rs9804659 and the best associated CNV marker (minimal P-value in the family-based discovery sample; Table 3). Thus CNVR 11q11 is not that well tagged by rs9804659, which might explain why the P-value in the replication is an order of magnitude lower than in the discovery sample. Table 3.Results of the replication analyses for CNVR-tagging SNPs in 365 independent case-parents obesity trios.CNVR no.ChromosomeStart – stop CNVR in BPTagging SNPPosition of tagging SNP in BPMax. squared correlation (r2), SNP – CNVR (tagging direction)aSquared correlation (r2), SNP – CNV marker with min. pfamily-based (tagging direction)bSNP obesity effect allelecSNP other alleleFrequency of obesity effect
