Table 2.
Performance evaluation based on TP53 benchmark data
| Methods | Missing | PPV (%) | NPV (%) | Specificity (%) | FPR (%) | Sensitivity (%) | FNR (%) | Accuracy (%) | MCC | AUC | hser-AUC | hspr-AUC |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Class one: function prediction methods | ||||||||||||
| FATHMM | 0 | 47.04 | NA | 0.00 | 100.00 | 100.00 | 0.00 | 47.04 | NA | 0.877 | 0.612 | 0.638 |
| fitCons | 0 | 47.04 | NA | 0.00 | 100.00 | 100.00 | 0.00 | 47.04 | NA | 0.521 | 0.521 | 0.501 |
| LRT | 29 | 70.38 | 82.73 | 66.67 | 33.33 | 85.05 | 14.95 | 75.53 | 0.524 | 0.731 | 0.556 | 0.513 |
| MutationAssessor | 1 | 64.49 | 92.56 | 53.36 | 46.64 | 95.18 | 4.82 | 73.05 | 0.526 | 0.880 | 0.634 | 0.674 |
| MutationTaster | 0 | 60.92 | 90.81 | 46.00 | 54.00 | 94.76 | 5.24 | 68.93 | 0.459 | 0.568 | 0.538 | 0.502 |
| PolyPhen2-HDIV | 0 | 58.48 | 93.30 | 38.92 | 61.08 | 96.86 | 3.14 | 66.17 | 0.430 | 0.865 | 0.655 | 0.556 |
| PolyPhen2-HVAR | 0 | 61.01 | 92.45 | 45.62 | 54.38 | 95.81 | 4.19 | 69.23 | 0.471 | 0.883 | 0.665 | 0.595 |
| PROVEAN | 0 | 76.64 | 87.77 | 76.16 | 23.84 | 88.05 | 11.95 | 81.76 | 0.643 | 0.898 | 0.688 | 0.666 |
| SIFT | 0 | 67.68 | 90.78 | 60.52 | 39.48 | 93.08 | 6.92 | 75.84 | 0.560 | 0.879 | 0.691 | 0.606 |
| VEST3 | 0 | 65.61 | 92.86 | 55.68 | 44.32 | 95.18 | 4.82 | 74.26 | 0.545 | 0.912 | 0.663 | 0.727 |
| Class two: conservation methods | ||||||||||||
| GERP++ | 0 | 57.67 | 81.23 | 41.90 | 58.10 | 89.10 | 10.90 | 64.10 | 0.347 | 0.732 | 0.577 | 0.541 |
| phastCons | 0 | 67.43 | 78.92 | 65.55 | 34.45 | 80.29 | 19.71 | 72.49 | 0.461 | 0.751 | 0.591 | 0.517 |
| phyloP | 0 | 64.73 | 78.69 | 60.52 | 39.48 | 81.55 | 18.45 | 70.41 | 0.427 | 0.802 | 0.561 | 0.584 |
| SiPhy | 4 | 65.45 | 66.84 | 71.48 | 28.52 | 60.38 | 39.62 | 66.24 | 0.321 | 0.731 | 0.537 | 0.531 |
| Class three: ensemble methods | ||||||||||||
| CADD | 0 | 63.71 | 92.08 | 51.96 | 48.04 | 94.97 | 5.03 | 72.19 | 0.512 | 0.841 | 0.679 | 0.577 |
| DANN | 0 | 64.78 | 76.85 | 61.82 | 38.18 | 79.04 | 20.96 | 69.92 | 0.412 | 0.752 | 0.638 | 0.520 |
| Eigen | 0 | 69.14 | 85.78 | 65.18 | 34.82 | 87.84 | 12.16 | 75.84 | 0.540 | 0.849 | 0.632 | 0.618 |
| FATHMM-MKL | 0 | 57.54 | 84.86 | 39.66 | 60.34 | 92.03 | 7.97 | 64.30 | 0.367 | 0.804 | 0.601 | 0.539 |
| GenoCanyon | 0 | 69.34 | 64.66 | 80.07 | 19.93 | 50.73 | 49.27 | 66.27 | 0.324 | 0.679 | 0.501 | 0.520 |
| M-CAP | 11 | 47.49 | 100.00 | 0.95 | 99.05 | 100.00 | 0.00 | 47.76 | 0.067 | 0.803 | 0.614 | 0.560 |
| MetaLR | 0 | 47.94 | 100.00 | 3.54 | 96.46 | 100.00 | 0.00 | 48.92 | 0.130 | 0.898 | 0.628 | 0.694 |
| MetaSVM | 0 | 48.13 | 100.00 | 4.28 | 95.72 | 100.00 | 0.00 | 49.31 | 0.144 | 0.578 | 0.562 | 0.562 |
| REVEL | 0 | 56.20 | 94.54 | 32.22 | 67.78 | 97.90 | 2.10 | 63.12 | 0.391 | 0.901 | 0.663 | 0.664 |
AUC, area under the curve; FNR, false negative rate; FPR, false positive rate; hser-AUC, high-sensitivity regional area under the curve; hspr-AUC, high-specificity regional area under the curve; MCC, Mathew correlation coefficient; NA, not available; NPV, negative predictive value; PPV, positive predictive value. The IARC TP53 Database compiles TP53 mutation data that have been reported in the peer-reviewed literature. We obtained 1014 somatic missense mutations that have been reported in this database, including 477 non-functional mutations and 537 functional mutations.