Table 3 Performances of the models on the tenfold cross-validation.

From: Coupling complementary strategy to flexible graph neural network for quick discovery of coformer in diverse co-crystal materials

Model

TPR (%)

TNR (%)

BACC (%)

SVMc

99.11 (±0.41)

89.81 (±3.55)

94.46 (±1.85)

RFc

99.82 (±0.15)

87.05 (±3.87)

93.44 (±1.89)

DNN-desc

99.55 (±0.19)

89.11 (±2.42)

94.33 (±1.25)

DNN-FPb,30

98.57 (±0.46)

86.48 (±4.86)

92.52 (±2.37)

enn-s2sa,47

98.63 (±0.38)

89.90 (±4.98)

94.27 (±2.41)

Graph-CNNa,48

98.94 (±0.39)

87.20 (±3.33)

93.07 (±1.60)

GCNa,30

98.98 (±0.43)

87.64 (±3.47)

93.31 (±1.76)

CCGNetd

99.82 (±0.14)

97.26 (±1.61)

98.54 (±0.79)

  1. aModel input is the molecular graph.
  2. bModel input is ECFP4.
  3. cModel input is the twelve molecular descriptors.
  4. dModel input is a combination of the molecular graph and the twelve molecular descriptors.
Back to article page