We imputed the simulated GWAS dataset from each reference panel using IMPUTE2 and Beagle. We ran Beagle on the same (default) settings as in the cross-validations, and we ran IMPUTE2 on two settings for each panel: khap = 500 and khap = N, where N is the number of haplotypes in the reference panel. As in the cross-validations, we fixed the IMPUTE2 phasing parameter at k = 80. Table 2 shows the single-processor running times and random-access memory (RAM) requirements of each program. We obtained these benchmarks from a single computer with 148 GB of RAM and a 2.4 GHz Intel Xeon processor.
