In calculating power, as thus defined, we simulate data under the assumption that a particular allele is causal and then look to see whether any SNPs on the respective genotyping chip, within a large region around the causal SNP attain the specified significance level. In ignoring the SNPs on the chip elsewhere in the genome, this approximation will underestimate the probability of there being a SNP meeting the significance threshold, but at the very low threshold, the probability of there being a SNP elsewhere in the genome meeting the threshold is extremely small, so that effect of this approximation will be minimal and our power calculations based on only on SNPs within the 1Mb region containing the causal SNP will be very close to the true values.
