In addition to what GWAS has taught us about genetic effect sizes, GWAS has also been informative as to the likelihood that a hypothesized candidate gene will be associated with the hypothesized outcome. Robust and replicable GWAS signals (e.g., CACNA1C for Schizophrenia; Ripke et al., 2013 and Bipolar Disorder; Ruderfer et al., 2013) have tended to be distinct from those that were routinely hypothesized in candidate gene studies (e.g., COMT, MAOA; Craddock, Dave, & Greening, 2001) casting considerable doubt regarding the burden of a priori evidence for these selections. With rare exceptions (e.g., rs16969968 in the CHRNA5-CHRNA3-CHRNB4 cluster, associated at p < 10–70 across several GWAS was initially posited as a candidate gene; Tobacco_and_Genetics_Consortium, 2010), widely studied candidate genes were not found to be significant when studied systematically across the backdrop of the genome in well-powered studies (F. J. Bosker et al., 2011; Collins et al., 2012; Lasky-Su et al., 2008). In fact, some of the findings to emerge from GWAS studies in other areas, such as Crohn’s disease, have suggested new pathways that were not previously suspected to play
