Despite being well-powered to confirm the majority of true associations, and identifying GSVs at similar overall frequencies to the original reports, only one reported association was confirmed using a case-control approach. We also conducted tests for association with BMI as a quantitative trait, for those loci at which GSVs were identified sufficiently frequently in our population cohort (for which there was no prior ascertainment on the basis of obesity). Of 3 GSVs present in >0.1% of subjects, 2 showed association with changes in BMI at or near nominal significance. Duplications of a region lying between the KIF2B and TOM1L1 genes showed marginal association with increased BMI, consistent with the original report [16]; there is at present no readily apparent functional basis for this putative association. Intriguingly, the second nominally-significant association was between deletions within the FOXP2 gene and decreased BMI, an effect in the opposite direction to that previously reported for the locus. This apparent directional inconsistency is likely to be due to the influence of GSVs that are appreciably larger than those previously reported, suggesting that the different variants
