best imputation results were achieved by running MACH twice, once with the HapMap CEU reference panel and a second time with the YRI reference panel, and then combining the results. In this way, we can capitalize on the higher accuracy of the YRI reference panel while also maintaining access to SNPs present only in the CEU reference panel. Methods to combine imputation results require more investigation. For this paper, we merged the separately generated data sets, giving preference to YRI calls when there were genotypes inconsistencies between the two data sets. Another possibility is to leverage local admixture estimates on a per-individual basis in order to determine which reference panel is the more appropriate choice.
