A striking feature of Figures 2 and 3 is that substantial differences in coverage between different chips do not translate into big differences in power. Put another way, coverage is often a poor surrogate for power. As an example, the coverage in the CEU HapMap population (r 2≥0.8) provided by the Affymetrix 500 k and Illumina 610 k chips are 65% and 87% respectively, a difference of 22%. On the other hand, the difference in power e.g. for relative risk 1.5 and 1500 cases and controls, is only 7% (66% and 73% respectively).
