Skip to main content

Table 4 Specific region of chemical space training and validation distribution

From: Feature combination networks for the interpretation of statistical machine learning models: application to Ames mutagenicity

Label

Feature

Training

Validation

  

Count

Active biasa

Count

Active bias

a

Aromatic amine (primary)

573

0.72

117

0.67

b

Aromatic amine (secondary)

113

0.61

28

0.61

c

Aromatic amine (tertiary)

168

0.60

38

0.63

d

Aromatic nitro

736

0.85

206

0.81

--b

Aziridine

39

0.95

13

1.00

e

Epoxide

248

0.75

62

0.61

f

Carboxylic acid

425

0.29

109

0.32

g

Aliphatic halogen

534

0.65

149

0.62

h

Bay-region polycylic hydrocarbon

190

0.86

39

0.87

  1. a = % of compounds in set with active experimental class, b No negative examples in the validation set.