Table 2.
Dataset | Antiviral drug | Target protein | Antiviral-protein interaction pair |
---|---|---|---|
Training set | 35 | 34 | 102c |
Validation seta | 61 | 47 | 182c,d |
Prediction set | 96 | 9b | 864 |
aValidation set consisted of U.S.FDA-approved antiviral drugs and these drugs’ binding target proteins.
b9 proteins of HPV-16.
cRatio of positive instance to negative instance was 1:1.
dNumber of validation set was greater than that of training set because (1) more FDA-approved antivirals were desired for validating the real-world application value of our machine learning models; (2) generalization performance of machine learning models could be reflected using smaller training set but larger validation set.