Figures 2 and 3 show the results of power calculation using the distribution of diversity in both the HapMap CEU and YRI populations. The results show that the increased ancestral recombination leads to a loss of power and coverage across all chips for a range of study designs. The difference between the power available from commercial genotyping chips and that achievable by exhaustively assaying all SNPs shows that increasing marker density may yield a better return than a similar approach in non-African populations. The Illumina 650 k chip, with the YRI fill-in illustrates these potential benefits, showing a marked increase in power over the 610 k. However the performance of the Illumina 300 k chip, designed using the CEU HapMap, falls below the Affymetrix 500 k when genetic diversity is modelled on the YRI HapMap panel.
