Table 1 Evaluation performance on the BindingDB_AIBind dataset under the four scenarios

From: ColdstartCPI: Induced-fit theory-guided DTI predictive model with improved generalization performance

Settings

Methods

Accuracy

Precision

Recall

F1 score

AUC

AUPR

S1

DNN

0.811 (0.007)

0.786 (0.038)

0.713 (0.040)

0.746 (0.007)

0.857 (0.003)

0.830 (0.002)

DeepConv-DTI

0.875 (0.005)

0.855 (0.013)

0.817 (0.012)

0.836 (0.005)

0.937 (0.004)

0.923 (0.004)

GraphDTA

0.849 (0.007)

0.801 (0.013)

0.814 (0.008)

0.807 (0.008)

0.910 (0.003)

0.882 (0.004)

MolTrans

0.873 (0.008)

0.864 (0.028)

0.802 (0.021)

0.831 (0.008)

0.929 (0.005)

0.915 (0.006)

ML-DTI

0.874 (0.007)

0.861 (0.018)

0.807 (0.015)

0.833 (0.008)

0.937 (0.005)

0.923 (0.005)

HyperAttentionDTI

0.889 (0.003)

0.847 (0.007)

0.873 (0.008)

0.860 (0.004)

0.950 (0.003)

0.938 (0.003)

DrugBAN

0.881 (0.004)

0.860 (0.008)

0.832 (0.016)

0.846 (0.007)

0.937 (0.003)

0.924 (0.004)

AI-Bind

0.892 (0.004)

0.866 (0.001)

0.854 (0.001)

0.860 (0.006)

0.947 (0.003)

0.936 (0.004)

ours

0.920 (0.004)

0.897 (0.001)

0.896(0.001)

0.897(0.004)

0.970 (0.001)

0.962(0.002)

S2

DNN

0.766 (0.019)

0.769 (0.040)

0.584 (0.078)

0.658 (0.044)

0.789 (0.019)

0.746 (0.008)

DeepConv-DTI

0.833 (0.011)

0.814 (0.015)

0.742 (0.022)

0.776 (0.011)

0.896 (0.011)

0.872 (0.008)

GraphDTA

0.828 (0.016)

0.793 (0.008)

0.756 (0.032)

0.774 (0.018)

0.882 (0.014)

0.844 (0.014)

MolTrans

0.824 (0.010)

0.8400 (0.009)

0.678 (0.027)

0.750 (0.016)

0.877 (0.009)

0.852 (0.012)

ML-DTI

0.824 (0.011)

0.810 (0.027)

0.721 (0.044)

0.761 (0.012)

0.893 (0.007)

0.864 (0.006)

HyperAttentionDTI

0.837 (0.011)

0.794 (0.028)

0.787 (0.021)

0.790 (0.013)

0.903 (0.009)

0.877 (0.010)

DrugBAN

0.850 (0.010)

0.840 (0.008)

0.749 (0.029)

0.792 (0.016)

0.906 (0.005)

0.883 (0.011)

DrugBAN_CDAN

0.846 (0.010)

0.834 (0.020)

0.758 (0.030)

0.794 (0.012)

0.902 (0.011)

0.882 (0.009)

AI-Bind

0.846 (0.010)

0.837 (0.001)

0.752 (0.001)

0.797 (0.010)

0.909 (0.008)

0.889 (0.006)

ours

0.876(0.011)

0.869(0.001)

0.802(0.001)

0.834(0.010)

0.936(0.009)

0.918(0.008)

S3

DNN

0.634 (0.021)

0.590 (0.066)

0.217 (0.042)

0.313 (0.043)

0.592 (0.018)

0.515 (0.035)

DeepConv-DTI

0.732 (0.018)

0.747 (0.033)

0.472 (0.022)

0.577 (0.014)

0.766 (0.006)

0.718 (0.026)

GraphDTA

0.649 (0.027)

0.577 (0.025)

0.355 (0.054)

0.437 (0.041)

0.643 (0.020)

0.561 (0.020)

MolTrans

0.716 (0.027)

0.764 (0.029)

0.392 (0.027)

0.517 (0.019)

0.738 (0.011)

0.688 (0.016)

ML-DTI

0.753 (0.015)

0.791 (0.046)

0.501 (0.032)

0.611 (0.015)

0.790 (0.005)

0.750 (0.025)

HyperAttentionDTI

0.754 (0.029)

0.767 (0.029)

0.536 (0.055)

0.628 (0.034)

0.796 (0.024)

0.754 (0.017)

DrugBAN

0.705 (0.038)

0.759 (0.028)

0.354 (0.046)

0.480 (0.041)

0.731 (0.022)

0.675 (0.007)

DrugBAN_CDAN

0.764 (0.014)

0.748 (0.058)

0.594 (0.023)

0.661 (0.025)

0.791 (0.020)

0.766 (0.036)

AI-Bind

0.807 (0.007)

0.811 (0.001)

0.652 (0.001)

0.722 (0.026)

0.863 (0.010)

0.832 (0.025)

ours

0.852(0.010)

0.848(0.001)

0.751(0.001

0.796 (0.025)

0.914(0.011)

0.891(0.024)

S4

DNN

0.565 (0.039)

0.404 (0.026)

0.236 (0.151)

0.273 (0.109)

0.513 (0.008)

0.399 (0.016)

DeepConv-DTI

0.624 (0.006)

0.528 (0.022)

0.345 (0.052)

0.416 (0.040)

0.626 (0.009)

0.514 (0.030)

GraphDTA

0.615 (0.009)

0.508 (0.026)

0.293 (0.048)

0.370 (0.044)

0.591 (0.013)

0.486 (0.033)

MolTrans

0.635 (0.010)

0.581 (0.031)

0.236 (0.022)

0.334 (0.023)

0.628 (0.010)

0.517 (0.019)

ML-DTI

0.663 (0.015)

0.613 (0.028)

0.381 (0.065)

0.466 (0.051)

0.682 (0.016)

0.579 (0.024)

HyperAttentionDTI

0.667 (0.009)

0.610 (0.025)

0.414 (0.033)

0.491 (0.015)

0.694 (0.014)

0.586 (0.011)

DrugBAN

0.639 (0.004)

0.607 (0.039)

0.243 (0.037)

0.346 (0.039)

0.637 (0.005)

0.534 (0.028)

DrugBAN_CDAN

0.687 (0.014)

0.642 (0.021)

0.450 (0.039)

0.528 (0.031)

0.694 (0.021)

0.642 (0.028)

AI-Bind

0.727 (0.007)

0.710 (0.001)

0.508 (0.001)

0.592 (0.008)

0.771 (0.008)

0.708 (0.012)

ours

0.773(0.005)

0.777(0.001)

0.586 (0.001)

0.668 (0.019)

0.839(0.009)

0.785(0.020)

  1. S1, S2, S3, and S4 represent the warm, compound cold, protein cold, and blind starts, respectively. The best results for each metric are highlighted in bold and the second-best results are underlined.