Table 7.
GATK | GATKcan (s.e.) | Mutect | VarScan 2 | ||
---|---|---|---|---|---|
trained by | |||||
10 tumors | 44 tumors | ||||
TPR§ | 89.9 | 98.7 (0.4) | 98.9 (0.3) | — | — |
cFPR | 64.1 | 4.4 (0.3) | 4.6 (0.2) | — | — |
precision | 27.6 | 85.9 (0.9) | 85.3 (0.6) | — | — |
MCC | 22.9 | 89.8 (0.7) | 89.5 (0.5) | — | — |
TPR†§ | 98.5 | 99.2 (0.3) | 99.2 (0.3) | 99.4 | 99.3 |
cFPR† | 76.6 | 8.8 (1.2) | 8.7 (1.1) | 57.0 | 47.9 |
precision† | 80.3 | 97.3 (0.3) | 97.3 (0.3) | 84.7 | 86.8 |
MCC† | 37.3 | 92.4 (0.3) | 92.5 (0.2) | 58.4 | 65.3 |
Next, the four algorithms identified mutations from 1,784 somatic variants, on which TPR†, cFPR†, precision† and MCC† were computed. §The unit of all performance measures and their s.e.’s are %.