We simulated Scenario B by modifying the WTCCC 58 C dataset as follows: First, we integrated the genotypes from the two SNP chips for the 1,377 shared 58 C individuals (see Text S1 for details), yielding a consensus set of 44,875 SNPs. Next, we split the 58 C samples into two groups: a diploid reference panel of 918 individuals (2/3 of the dataset) and a study sample of 459 individuals. To complete the reference panel, we added 120 haplotypes from the HapMap Phase II CEU data. We then created two Scenario B study sample datasets by masking the genotypes of SNPs unique to each chip in turn; there were 18,489 such SNPs on the Affymetrix chip and 22,219 such SNPs on the Illumina chip.
