Table 6.
Comparison of Computation Times (in Hours) for Different Methods[Note]
|
Missing-Genotype Estimation |
||||||
| HapMapChromosome 7 |
HapMapChromosome 22 |
|||||
| Method | SeattleSNPs (10%) |
10% | 25% | 10% | 25% |
Haplotype Inference for X Chromosome (7%) |
| fastPHASE | 1.3 | 9 | 8 | 5 | 5 | 2 |
| PHASE version 2 | 29.3 | NA | NA | 323 | 720 | 151 |
| GERBIL | .6 | 29 | 32 | 10 | 8 | 2 |
| HaploBlock | 160.1 | NA | NA | NA | NA | 265 |
Note.— Each method was applied on a 3-GHz Xeon processor with 1 GB of memory. The SeattleSNPs column is for analyses of all 47 individuals together. For missing-genotype estimation, fastPHASE was applied to the data without inference of haplotypes. Calculation times for HaploBlock with the SeattleSNPs data and for PHASE version 2 with HapMap chromosome 22 data are based on extrapolation from a subset of the data sets. Approximate percentages of missing data are in parentheses. Bold italics indicate the fastest method(s) for each data set.