In contrast to common SNPs, most low-frequency SNPs are not shared across diverged populations because they have either arisen relatively recently or their frequencies have been influenced by population history (e.g., the out-of-Africa expansion or natural selection). Since the synthetic associations proposed by Dickson et al. are created by low-frequency variants, they are therefore less likely to be shared among diverged populations. The Crohn's disease synthetic association at NOD2 is again illustrative, as it is restricted to populations of European or Jewish descent [27]–[29]. The three rare causal mutations are not observed in Asian populations, so the common variant (which is observed at similar frequency to that seen in individuals of European descent) is not associated with Crohn's disease in individuals of Asian descent. In contrast, many disease loci initially identified via GWAS have been widely replicated across divergent populations, supporting the hypothesis that the causal mutations are, in fact, common [30]–[32]. Some population-specific GWAS signals do exist [6], but caution is required in the interpretation of small-scale, population-specific replication studies of loci originally identified in meta-analyses involving thousands of
