search to include an extra four such loci but did not identify any further overlaps. The remaining 47 type 1 diabetes risk loci [20] therefore show no evidence of linkage. Indeed, for complex diseases more generally there has been little overlap in the genomic regions identified using linkage and association [21]. Thus, either synthetic association is confined to a relatively narrow set of genetic models that are detectable only by GWAS or it does not account for many known associations.
