Using MACH, we also found that the HapMap phase III YRI reference panel slightly outperformed the HapMap phase III ASW reference panel. A priori, potential advantages of the ASW reference panel include more representative allele frequencies (the average estimated FST is 0.026 between the ASW reference panel and our study sample compared to 0.0295 between the YRI reference panel and our study sample) and linkage disequilibrium patterns resulting from admixture. On the other hand, potential disadvantages of the ASW reference panel include 2.6-fold less dense SNP sampling and 2.3-fold fewer founders compared to the combined phase II+III YRI data. It is indeed possible that ASW may perform equally to or better than YRI after eliminating these two differences in sampling.
