However, with underlying polygenic genetic architectures, effects of only modest magnitude are likely to be identified in many single samples of practical size. We and others have developed “nontemplate” GWA analyses to address highly heritable complex phenotypes for which there is little evidence for many genes of major effect. These analyses have focused on identification of nominally significant case vs control allele frequency differences at several nearby SNP markers in multiple independent samples. Identifying “clustered” positive findings at several nearby SNPs and finding clustered positive results in several independent samples provide some of the best available controls for technical errors and for the large numbers of repeated comparisons that are fundamental to GWA. There is no consensus concerning criteria for declaring “replication” of GWA results in the absence of genome wide significance for the same SNP with the same phase of association in multiple independent samples [18]–[23]. Several considerations have prompted differing approaches to 1) combining and comparing GWA datasets and 2) declaring that association between sets of nearby SNPs and a complex disorder is “replicated” in the absence of
