Table 5. The performance of different servers for the independent dataset.
| Server Type | Data for server construction | Server | Available | Mean AUC |
| Structured-based | Bound structure | DiscoTope | http://www.cbs.dtu.dk/services/DiscoTope/ | 0.579 |
| SEPPA | http://lifecenter.sgst.cn/seppa/ | 0.589 | ||
| EPITOPIA | http://epitopia.tau.ac.il/ | 0.572 | ||
| BPredictor | http://code.google.com/p/my-project-bpredictor/ | 0.587 | ||
| Unbound structure | EPCES | http://sysbio.unl.edu/EPCES/ | 0.569 | |
| EPSVR | http://sysbio.unl.edu/EPSVR/ | 0.606 | ||
| Sequence-based | Sequence dataset | CBTOPE | http://www.imtech.res.in/raghava/cbtope/ | 0.607 |
| Bound sequence | Our model1 | 0.600 | ||
| Unbound sequence | Our model2 | 0.601 | ||
| CBTOPE dataset | Our model3 | 0.632 |
Our model1 is constructed on the sequence dataset compiled from the bound structures; our model2 is constructed on the sequence dataset compiled from the unbound structures; our model3 is constructed on the dataset, which was used for CBTOPE.