We simulated data for twenty-two one megabase regions chosen at random, one from each autosome. To ensure that the regions used to approximate genome-wide power were representative of the genome at large we their SNP density. Figure S5 plots the distribution of inter-SNP distances within the 22 analysis regions and across the whole genome for three of the genotyping chips analyzed. The close match between the distribution, both on the physical scale and in terms of genetic distance suggests that our results are insensitive to the regions we chose to simulate, and can be used to make comparisons of genotyping chips genome-wide.
