Next, there is a clear gain in power for tests based on rare variants identified through re-sequencing over analyses of SNPs or low-frequency variants present on the GWA chip. The greatest gains are observed in the presence of substantial allelic heterogeneity (Fig. 1a), where rare causal loci are less likely to be captured by SNPs as a result of linkage disequilibrium [The International HapMap Consortium, 2005; Zeggini et al., 2005]. However, the differences in power between the tests are less noticeable when there is less allelic heterogeneity (Fig. 1b). Our results also confirm previous findings that haplotype-based analyses of SNPs have greater power to detect rare variant associations than single-locus tests, unless there is substantial allelic heterogeneity [Morris and Kaplan, 2002]. Finally, low-frequency variants (MAF<5%) on GWA chips are too scarce to detect accumulations of minor alleles, and thus RVT1 and RVT2 have minimal power to identify rare variant associations with this type of data.
