We took data from the 10 ENCODE regions [ENCODE Project Consortium, 2004] within the Caucasian (CEU) analysis panel of HapMap II [International HapMap Consortium, 2007], which have undergone SNP ascertainment by resequencing. These regions therefore show a fuller spectrum of SNPs than are represented in the HapMap data at large, and haplotypes are expected to be accurate due to the trio design of the HapMap panels [International HapMap Consortium, 2005]. We used the HAPGEN software package [Spencer et al., 2009] to produce a population of 100,000 haplotypes based on the empirical LD patterns in HapMap II. This haplotype panel served as the base for our GWAS simulations.
