Our results provide statistically significant evidence of a high burden of large CNVs (>500kb) in TS, thereby supporting the proposal for an involvement of rare CNVs in various neurodevelopmental disorders, including TS, and their possible aetiological overlap [12], [13], [26]–[28]. We also find suggestive evidence for the involvement of rearrangements specifically affecting the NRXN1 and COL8A1 genes. In the aggregated data (Table 4) we find a nominally significant association of COL8A1 and NRXN1 rearrangements with TS (p-values of 0.004 and 0.03 respectively). Due to the limited sample size, these p-values would not reach significance accounting for multiple testing. Data from the Database of Genomic Variants further supported the notion that the variants observed here are not part of the spectrum of common population polymorphisms. When considering the trio data, the lack of a straightforward co-segregation between the structural variants observed in our study and the TS phenotype implies the involvement of further predisposing loci in the aetiology of TS; however, this is not unexpected for such a phenotypically and genetically complex condition and does not conflict with a role for NRXN1 or COL8A1 in TS predisposition. Overall, our results strongly warrant further investigation of these two genes in TS.
