Table 2 Comparison results of EviDTI and baselines on the KIBA dataset
From: Evidential deep learning-based drug-target interaction prediction
KIBA | Accuracy (std) | Recall (std) | Precision (std) | MCC (std) | F1 score (std) | AUC (std) | AUPR (std) |
|---|---|---|---|---|---|---|---|
RF | 88.41 (0.34) | 64.84 (2.48) | 70.57 (2.30) | 63.21 (1.67) | 67.59 (1.37) | 91.05 (0.38) | 76.05 (1.04) |
NB | 52.92 (0.83) | 71.72 (1.77) | 24.22 (1.40) | 24.56 (3.24) | 36.21 (1.47) | 65.92 (0.78) | 35.51 (0.94) |
SVM | 81.72 (0.46) | 3.87 (0.76) | 66.40 (2.91) | 5.24 (1.34) | 7.32 (0.92) | 73.27(0.56) | 43.38 (1.60) |
DeepConv-DTI | 89.05 (0.32) | 68.87 (0.43) | 80.52 (0.34) | 70.21 (0.28) | 74.24 (0.63) | 90.67 (0.17) | 80.42 (0.62) |
GraphDTA | 88.92 (0.11) | 59.42 (3.23) | 77.53 (2.03) | 67.51 (0.52) | 71.24 (0.81) | 91.41 (0.12) | 77.61 (0.74) |
MolTrans | 88.91 (0.16) | 73.53 (0.23) | 70.42 (0.34) | 67.62 (0.22) | 71.94 (0.16) | 92.32 (0.33) | 79.49 (0.64) |
HyperAttention | 89.33 (0.12) | 79.63 (0.24) | 68.93 (0.18) | 70.12 (0.22) | 74.21 (0.35) | 93.51 (0.17) | 81.41 (0.23) |
TransformerCPI | 87.01 (0.18) | 63.12 (0.32) | 66.93 (0.34) | 66.2 (0.26) | 69.12 (0.23) | 88.81 (0.18) | 70.81 (0.16) |
GraphormerDTI | 86.54 (0.85) | 85.40 (1.68) | 60.63 (1.84) | 64.03 (1.10) | 70.87 (0.89) | 92.58 (0.25) | 81.95 (0.91) |
AIGO-DTI | 90.68 (0.32) | 73.65 (0.77) | 76.77 (1.12) | 69.46 (0.97) | 75.17 (0.77) | 93.06 (0.26) | 83.91 (0.77) |
DLM-DTI | 88.25 (0.25) | 68.10 (2.15) | 69.86 (0.90) | 61.72 (0.78) | 68.94 (0.87) | 91.14 (0.39) | 75.63 (0.41) |
EviDTI | 91.27 (0.07) | 70.77 (1.69) | 80.90 (1.32) | 70.44 (0.78) | 75.48 (0.83) | 93.57 (0.35) | 83.03 (0.98) |
