For additional array-based replication analyses of the 20 discovered CNVRs in a sample of 281 independent obesity trios, intensity values for each involved marker were again extracted with the R-package ‘affxparser’. As for the family-based discovery GWAS sample, the CNV-FBAT methodology was applied to the raw intensities afterwards. Finally, one-sided P-values were calculated by taking into account the estimated CNVR effect direction as based on the family-based discovery GWAS sample.