When there is ample evidence for the involvement of a given gene or system but little prior knowledge regarding specific polymorphisms within it, methods have been introduced that average across associations for all variants within the gene/system, typically regardless of function, to create a summary statistic (Holmans, 2010; Li, Gui, Kwan, & Sham, 2011; Liu et al., 2010; Neale & Sham, 2004; Ramanan et al., 2012; Wang et al., 2007, 2010). Numerous approaches have been proposed to mine GWAS data in this manner. For instance, the Versatile Gene-Based Association Analysis (VEGAS; Liu et al., 2010) software package utilizes p-values for each SNP from a typical GWAS, assigns them to one of 17,787 autosomal genes, and creates a sum statistic representing gene-based association whose empirical p-value is calculated via simulations. This approach has successfully identified genes associated with cannabis dependence when single-SNP GWAS have failed (Agrawal et al., 2014).
