Some of the genetic risk for schizophrenia is not mediated via common SNPs but by CNVs and other rare variants. CNVs are lengths of DNA (of the order of a million nucleotides) which are either deleted or duplicated, but which are too small to have been seen using conventional karyotyping methods. Recent microarray and other technologies show that CNVs are a normal feature of the genome, but also that CNVs which affect particular genomic regions are associated with an increased risk of schizophrenia. Eight such genomic regions are well established (Mowry and Gratten, 2013). Most encompass multiple genes, although two affect a single gene: 2p16.3 deletion (NRXN1; neurexin 1), and 7q36.3 duplication (VIPR2; vasoactive intestinal peptide receptor 2). Compared with schizophrenia-associated SNPs, each CNV is penetrant (Kirov et al., 2014) and confers a significantly increased risk of illness (odds ratios for several CNVs exceed 8), but each CNV is extremely rare, except for hemideletion of 22q11 (velocardiofacial syndrome; Schneider et al., 2014). The contribution that CNVs make overall to the aetiology of schizophrenia is unknown; a recent population-based study estimated
