Table 1.
Switch error (%) obtained by the tools when applied on chromosomes 1, 6, and 17
| Approach | Tool | Chr1 | Chr6 | Chr17 |
|---|---|---|---|---|
| Individual | BEAGLE | 1.39 | 1.12 | 2.03 |
| SHAPEIT2 | 1.33 | 1.20 | 2.06 | |
| EAGLE2 | 1.36 | 1.6 | 1.90 | |
| HAPI-UR | 2.14 | 1.81 | 3.00 | |
| IMPUTE | 2.83 | 2.46 | 4.30 | |
| fastPHASE | 4.10 | 3.43 | 5.32 | |
| MaCH | - | - | 4.10 | |
| Consensus | SHAPEIT2, EAGLE2 and BEAGLE | 1.14 | 0.98 | 1.68 |
| SHAPEIT2, EAGLE2, BEAGLE, IMPUTE and HAPI-UR | 1.2 | 1.04 | 1.76 | |
| SHAPEIT2, EAGLE2 and HAPI-UR | 1.24 | 1.08 | 1.79 | |
| SHAPEIT2, BEAGLE and fastPHASE | 1.37 | 1.14 | 2.06 | |
| SHAPEIT2, EAGLE2, IMPUTE, fastPHASE and HAPI-UR | 1.41 | 1.19 | 2.1 | |
| EAGLE2, IMPUTE and HAPI-UR | 1.48 | 1.27 | 2.16 | |
| SHAPEIT2, fastPHASE and HAPI-UR | 1.66 | 1.43 | 2.41 | |
| IMPUTE, fastPHASE and HAPI-UR | 2.16 | 1.82 | 3.19 |
Numbers in bold are the minimum error obtained according to each approach while underlined numbers are the minimum error obtained in all applied tests. MaCH was applied only on chromosome 17 due to the intensive performance time needed to phase other chromosomes. The tools are sorted according to the average switch error. The top section of the table lists the performance of the tools individually, while the bottom half lists the performance of the consensus estimators based on different combinations of phasing methods