Table 2.
Comparison of prediction performance of EXIA2 web server and CRpred, POOL, ConSurf, and ResBoost.
| Competing method | CRpred | POOL1 | POOL2 | ConSurf3 | ResBoost3 |
|---|---|---|---|---|---|
| Benchmark datasets | EF fold | POOL160 | POOL160 | P100 | P100 |
| Recall (R) | 48.2 | 61.7 | 64.7 | 55.0 | 55.0 |
| Precision (P) | 17.0 | 18.1 | 19.1 | 5.0 | 17.0 |
| AUCROC4 | — | 0.907 | 0.925 | — | — |
| EXIA2 + PSSM5 | |||||
| Recall at equal P | 72.7 | 80.0 | 77.8 | 96.0 | 74.3 |
| Precision at equal R | 27.3 | 24.3 | 23.3 | 25.3 | 25.3 |
| AUCROC | 0.968 | 0.965 | 0.965 | 0.966 | 0.966 |
| EXIA26 | |||||
| Recall at equal P | 45.1 | 70.8 | 68.6 | 90.6 | 58.0 |
| Precision at equal R | 16.2 | 22.2 | 20.8 | 18.3 | 18.3 |
| AUCROC | 0.943 | 0.960 | 0.960 | 0.952 | 0.952 |
1Prediction results of POOL.
2Prediction results of POOL combined with evolutionary information.
3Prediction results published in [40]. Complete comparison of ROC and recall-precision curves is available in supplementary Figure S4.
4Some AUC values are not available in the publications.
5Prediction results of EXIA2 combined with PSI-Blast PSSM.
6Prediction results of EXIA2 without evolutionary information.