We assessed the power of commercially available chips via simulation. Each simulation assumed a particular SNP in HapMap was causative, with a given effect size and used the HAPGEN package to simulate case and control samples of different sizes. In the simulated data we then restrict attention to the genotypes at only the SNPs on the chip in question and ask whether analysis of these would yield a significant result for any of the SNPs on the chip. An estimate of power is obtained by repeating the simulation over a large number of putative disease SNPs across the genome and using the proportion of simulations in which we find a significant test statistic.
