The dearth of synthetic associations reported to date could be due to the low-frequency variants residing outside the resequenced interval [7]. For example, analysis of a GWAS of sickle cell anemia demonstrated that strong GWAS signals can be observed at large distances from causal alleles. The genome-wide significant (P<5×10−8) signal indeed spans multiple LD blocks (2.5 Mb in total), due to the fully penetrant causal (recessive) genotype, in stark contrast to the small increase in risk from typical GWAS hits. Even so, the most associated SNP (P = 1.1×10−136) is within 10 kb of the known functional variant in HBB [6]. The notion that GWAS signals are typically located close to underlying functional elements is further supported by their frequent proximity to candidate genes associated with related Mendelian conditions or identified by pathway analyses [24],[25]. For example, eight of ten proteins involved in the Th17-differentiation signaling pathway have been associated with one or more auto-inflammatory diseases [26]. Identifying so many candidate genes within the narrow intervals around GWAS signals suggests that causal variants are not routinely located megabases away from the most strongly associated common SNP.
