To derive CNVRs for replication using CNVR-tagging SNPs, we first focused on those estimated CNVRs which showed some evidence for association to obesity in the family-based GWAS sample (CNV-FBAT test; at least one marker with a two-sided asymptotic P-value ≤ 0.05). Taking into account whether this CNVR indicated gain or loss for the obesity association, we then calculated CNV marker-specific one-sided P-values for the tests in the case–control GWAS sample. For clarity (if not stated otherwise), all reported P-values are one-sided, nominal, i.e. unadjusted for multiple testing, and asymptotic.
