In light of these considerations, we decided to create two versions of our Scenario B dataset: one that includes the full set of HapMap SNPs, and one in which the HapMap dataset is restricted to SNPs that were genotyped on at least one of the chips (i.e., in which all SNPs in U1 have been removed). We used the latter dataset to broaden the range of methods that could be included in the comparison (at the cost of removing some of the complexity of Scenario B), and we used the former dataset to evaluate BEAGLE and IMPUTE v2 in a more realistic setting.
