a 10 Mb block that the most associated SNP may be positioned several Mb from the most distal causal variant it is coupled with. Even if associations from causal variants spanned several Mb for synthetic associations, to have an important contribution to disease variance we would expect to see associations concentrated in only a fraction of the genome. This will be a testable hypothesis as sample sizes increase; results to date do not support such a concentration of associations. Since the simulation parameters used by Dickson et al. generate a relatively small number of rather large loci, their results may generate a false sense of the relative importance of the contribution of synthetic associations to GWAS results observed to date.
