calls across the entire region were combined, and individuals with copy number changes (i.e. deletion or duplication, as appropriate) at over 50% of probes within the region were provisionally called as carrying a GSV; the presence of a full-length GSV in these individuals was then confirmed by manual inspection of the LRR and BAF data. This approach again minimises the potential for artefactual associations arising from different GSV call-rates in cases and controls (e.g. due to differences in DNA quality). Furthermore, for both adult and child case-control cohorts, any potential bias in GSV detection favoured an increased call rate in cases (higher mean number of GSV calls per subject) which would be expected to favour false positive associations and to mitigate against false negatives. All chromosomal coordinates are given according to genome build 36 (hg18).
