Undoubtedly, part of the missing heritability is explained by imperfect LD between the genotyped SNPs and causal variants, including rare causal variants and including multiple rare causal variants concentrated in relatively short genomic regions. Dickson and colleagues [4],[5] give six examples of known synthetic associations detectable in GWAS but generated by rare causal variants, providing compelling evidence for their existence. However, their examples likely represent the high end of effect size of rare causal variants (since in many instances the causal variants had been detected in the pre-GWAS era) and yet, in all examples, the RAF of the most associated tag SNPs (13 were listed across the examples) were less than 0.33. However, the majority of associations detected from GWAS are for very common SNPs (Table 3); although very common associated SNPs are not likely to be the causal variants, they are much more likely to tag causal variants of similar frequency and highly unlikely to represent synthetic associations with single or multiple rare causal variants. Thus, we conclude that synthetic associations created by rare variants are unlikely to explain
