Table 5.
The general performance test of the PredPhospho for two real data sets
| Specificitya | Prediction at the group level |
Specificity | Prediction at the family level |
||||||
|---|---|---|---|---|---|---|---|---|---|
| Data I |
Data II |
Data I |
Data II |
||||||
| Sn (%) | Sp (%) | Sn (%) | Sp (%) | Sn (%) | Sp (%) | Sn (%) | Sp (%) | ||
| No | 79.40 | 60.62 | 75.47 | 61.04 | No | 95.36 | 29.29 | 93.60 | 30.98 |
| 95% | 73.24 | 72.09 | 65.76 | 72.39 | 95% | 92.68 | 42.80 | 88.95 | 46.63 |
| 97% | 53.37 | 80.22 | 48.31 | 79.45 | 97% | 88.46 | 56.81 | 83.50 | 57.36 |
| 98% | 43.11 | 89.23 | 38.92 | 89.26 | 98% | 82.03 | 66.80 | 75.44 | 62.88 |
| 99% | 23.39 | 95.79 | 20.05 | 96.62 | 99% | 73.18 | 72.67 | 66.73 | 72.39 |
aOptions of the specificity. For example, ‘99%’ specificity option mean cutoff value is adjusted for each model to have 99% specificity, and ‘No’ specificity option means each model has default cutoff value without adjustment of specificity (See supplementary material).
Abbreviations: sensitivity, Sn; specificity, Sp.