detect it even if the best SNP on the chip only has r 2 = 0.5 or less. At the other extreme, for rare SNPs, unless the effect size is very large, power would be low even if the SNP had a perfect proxy on the chip. Thus even if these SNPs were well covered by one chip and completely missed by another they would not contribute to a difference in power between the chips because both chips would have power close to zero for them. The top row of Figure 4 shows the average power for SNPs in each LD bin. For the Affymetrix 500 K chip, there is a greater contribution to power from the sets of SNPs which are not well “covered”, than for the Illumina chip, and hance a smaller difference in power than in coverage.
