The existing findings account for only a minority of the heritability of schizophrenia, and there are various explanations for what accounts for the rest (Gibson, 2012; Lee et al., 2011; Manolio et al., 2009). Firstly, as noted, many more SNPs, and more CNVs and rare variants, will no doubt be found as exome- and genome-sequencing studies bear fruit. Secondly, it is likely that at least some, and possibly much, of the genetic risk reflects gene–gene interactions (epistasis; Mackay, 2014) rather than simply the cumulative effect of multiple independent genes (Phillips, 2008). There are preliminary clinical (Nicodemus et al., 2010), and experimental (Papaleo et al., 2014) data which support a role for epistasis in schizophrenia, but it has not been systematically investigated, and GWAS studies are not well suited to identify it. Thirdly, epigenetic factors (such as DNA methylation and histone modifications) may be involved (Daxinger and Whitelaw, 2012; Dempster et al., 2013), and contribute to gene–environment interactions, whereby part of the genetic risk for schizophrenia operates by altering sensitivity to environmental factors, such as obstetric complications or early use of
