The simulations of Dickson et al. (Box 1) show that for the genetic models they consider, SNPs with frequency typical of those represented on GWAS panels could tag single or multiple causal variants. They show the likely frequency of the most associated genotyped SNP (their Figure 5), which provides a benchmark for comparisons with empirical data. To understand the results of Dickson et al requires an understanding of linkage disequilibrium (Box 2) between rare and common variants. We first consider the situation of a synthetic association resulting from a single causal variant. In association studies, the genotyped SNP is unlikely to be the causal variant, but how likely is it that a GWAS reported association, the most associated common SNP in a region, tags rare causal variant(s)?
