Ascertainment according to broad criteria can give only limited enrichment of rare variants. Conversely, cohorts selected on the basis of pronounced phenotypes may be difficult to recruit, a potential problem not only for variant discovery but also for replication, especially where the discovery cohort includes individuals not normally recruited to other cohorts (e.g. those with DD). For instance, we were unable to attempt replication of GSVs that were originally identified in cases with severe obesity plus DD since very few of these were detected in our cohorts; some of these may reflect novel ‘syndromic’ forms of obesity whose replication will require analysis of additional ‘obesity plus’ cohorts. We also note that recruitment of cohorts of sufficient size by including subjects from a range of localities or ethnicities may introduce complications, since the majority of population groups, both within Europe and worldwide, have not yet been assessed for population specific GSVs. This also poses a problem for replication – the failure to detect some GSVs in our cohorts may reflect low frequencies in the populations from which they were drawn. Furthermore,
