Table 6 Prediction performance of models in the ablation study
From: PlantNh-Kcr: a deep learning model for predicting non-histone crotonylation sites in plants
Model | Sn (%) | Sp (%) | ACC (%) | F1Ë—score (%) | MCC (%) | AUC (%) |
|---|---|---|---|---|---|---|
Removing linear layer | 90.0 \(\pm\) 1.19 | 70.7 \(\pm\) 2.24 | 74.8 \(\pm\) 1.56 | 60.1 \(\pm\) 1.27 | 50.2 \(\pm\) 1.46 | 88.8 \(\pm\) 0.32 |
Removing CNN | 78.0 \(\pm\) 4.48 | 82.6 \(\pm\) 3.00 | 81.6 \(\pm\) 1.50 | 64.1 \(\pm\) 0.92 | 53.9 \(\pm\) 1.09 | 88.7 \(\pm\) 0.44 |
Removing MHSA | 79.2 \(\pm\) 2.33 | 84.5 \(\pm\) 1.45 | 83.4 \(\pm\) 0.70 | 66.7 \(\pm\) 0.51 | 57.3 \(\pm\) 0.64 | 89.4 \(\pm\) 0.44 |
Removing BiLSTM + MHSA | 82.1 \(\pm\) 1.08 | 82.2 \(\pm\) 0.80 | 82.1 \(\pm\) 0.43 | 66.0 \(\pm\) 0.33 | 56.5 \(\pm\) 0.38 | 89.6 \(\pm\) 0.07 |
PlantNh-Kcr | 81.1 \(\pm\) 3.23 | 83.3 \(\pm\) 2.09 | 82.8 \(\pm\) 0.99 | 66.5 \(\pm\) 0.50 | 57.2 \(\pm\) 0.50 | 89.9 \(\pm\) 0.19 |