Table 2.
The k-fold cross-validation performance of our method vs. DeepSuccinylSite (the data are percentages). DeepSuccinylSite a and DeepSuccinylSite b denote DeepSuccinylSite with one-hot encoding and embedding encoding. The highest value in each category is shown in italic.
| Method | Sn | Sp | Acc | MCC | Gmean | AUC | AUPR |
|---|---|---|---|---|---|---|---|
| DeepSuccinylSite a | 69.84 | 68.11 | 69.87 | 38.16 | 68.77 | 75.92 | 72.63 |
| DeepSuccinylSite b | 75.69 | 55.62 | 65.66 | 32.03 | 64.80 | 71.66 | 68.93 |
| Our Method (k = 4) | 63.69 | 77.48 | 76.32 | 26.11 | 69.76 | 79.36 | 24.28 |
| Our Method (k = 6) | 60.58 | 79.59 | 77.99 | 26.48 | 68.65 | 79.73 | 24.96 |
| Our Method (k = 8) | 67.26 | 75.71 | 74.99 | 26.80 | 71.03 | 79.94 | 25.33 |
| Our Method (k = 10) | 66.81 | 76.75 | 75.91 | 27.36 | 71.37 | 80.35 | 25.88 |