Skip to main content

Table 6 Performance of Models on New training and validation dataset built using MACCS fingerprints

From: DrugMint: a webserver for predicting and designing of drug-like molecules

Dataset

Sensitivity

Specificity

Accuracy

MCC

AUC

Model based on 159 MACCS keys after rm-useless

New train

90.25

89.40

89.65

0.77

0.95

Validation

90.37

87.21

88.14

0.77

0.95

Model based on Top-20 PCs

New train

85.70

88.65

87.78

0.72

0.94

Validation

81.85

87.21

85.62

0.67

0.92

Model based on 10 MACCS keys after CfsSubsetEval

New train

89.42

81.83

84.07

0.67

0.89

Validation

84.07

81.44

82.22

0.62

0.88