Our replication of the obesity association of deletions including SH2B1 and the finding of limited evidence to support 2 further associations, together with recent successes in other disorders including attention deficit hyperactivity disorder [43], demonstrate that analysis of carefully-selected cohorts has the potential to reveal novel, rare, causal GSVs. However, it is clear that there remains a need for an accepted foundation on which to base genome-wide searches for rare variants. In its absence, attempts to overcome the unavoidable lack of statistical power may lead to the adoption of methods whose effectiveness is not readily quantifiable. Thus, there is a danger that reported associations may include a large number of false positives. Similar caveats should perhaps also be attached to the growing number of studies investigating common GSVs [36], [37], [39]. Although careful experimental design and the inclusion of additional phenotypic and/or experimental data can help to limit this problem, our findings illustrate the urgent need for well-defined, robust statistical methods that are readily applicable to the search for causal, rare, genomic structural variants.
