Table 3.
Classifier | Level | Sp (%) | Sn (%) | ACC (%) | Pre (%) | MCC (%) |
---|---|---|---|---|---|---|
Phos-based | High | 94.8 | 70.0 | 88.3 | 83.8 | 69.0 |
Medium | 90.0 | 77.8 | 86.7 | 74.2 | 66.8 | |
Low | 85.0 | 82.0 | 84.2 | 66.9 | 63.2 | |
| ||||||
YinOYang ∗1 | ∖ | 75.5 | 42.7 | 68.9 | 67.5 | 32.7 |
| ||||||
O-GlcNAc-based | High | 94.9 | 50.4 | 83.0 | 78.8 | 63.3 |
Medium | 90.0 | 57.3 | 81.2 | 67.9 | 50.0 | |
Low | 85.2 | 70.2 | 81.0 | 63.3 | 53.5 | |
| ||||||
YinOYang ∗2 | ∖ | 71.2 | 42.7 | 56.8 | 59.6 | 14.8 |
For YinOYang ∗1 and YinOYang ∗2, we used the same method to evaluate the performance of YinOYang while the test sets were different. For YinOYang ∗1, the test dataset is the same with that used for phos-based classifier and, for YinOYang ∗2, the test dataset is the same with that used for O-GlcNAc-based classifier.