Skip to main content

Table 1 Characteristics of the studied benchmark data sets

From: Efficiency of different measures for defining the applicability domain of classification models

 

Descriptor type

No. of objects

No. of descriptors

Class ratio (%)

Musk2

Shape/conformation

6598

166

85/15

QSAR

DRAGON

1055

41

34/66

BBB

Chemical/physical properties

325

9

45/55

PGP

Binarized atom pairs

186

1522

42/58

CYP1A2

MACCS/MOE/E-State

7485

166/181/192

46/54

FXa

MACCS/MOE

435

166/181

36/64

Liver

MACCS/MOE

951

166/181

32/68

hERG

MACCS/MOE

561

166/181

62/38

Cancer

MACCS/MOE

7747

166/181

41/59

Ames

MACCS/MOE

6512

166/181

46/54