Table 2. Comparisons among 3 tools: CHASM, MutationTastor, and CanDrA.
CHASM | MutationTastor | CanDrA | ||||
Datasets | AUC | Recall | AUC | Recall | AUC | Recall |
GBM.S1 | 0.890 | 49/67 | 0.892 | 56/67 | 0.911 | 56/67 |
GBM.S2 | 0.923 | 44/67 | 0.909 | 48/67 | 0.941 | 48/67 |
OVC.S1 | 0.936 | 50/61 | 0.910 | 46/61 | 0.953 | 50/61 |
OVC.S2 | 0.940 | 50/61 | 0.910 | 40/61 | 0.953 | 47/61 |
Two metrics are compared: 1) The area under the receiver operator characteristics curve (AUC) for classifying drivers and passengers and 2) the number of drivers predicted by each of the tools vs. the total number of drivers (recall) in each of the 4 test datasets.