As in Scenario A, we re-ran all programs on a single Linux server to assess their computational burdens in Scenario B. We used the same server and analysis chunks as before; the average running times and memory requirements across these four 7.5 Mb regions of chromosome 10 are shown in Table 3. (These numbers were averaged across both the masked Affymetrix and masked Illumina datasets, so eight runs contributed to each table entry.) Numbers in parentheses refer to the full Scenario B dataset, and all other numbers refer to the restricted dataset. Note that IMPUTE v1 was run on a version of the dataset that did not include the diploid reference panel.
