Skip to main content

Table 9 Description of the important features in the consensus model when the cutoff is 50%

From: HobPre: accurate prediction of human oral bioavailability for small molecules

Descriptor category

Feature name

|shap value|

Description

Estate

SsOH

0.0239

Sum of sOH

TopoPSA

TopoPSA(NO)

0.0178

Topological polar surface area (use only nitrogen and oxygen)

Autocorrelation

ATSC0c

0.0172

Centered Moreau–Broto autocorrelation of lag 0 weighted by Gasteiger charge

Autocorrelation

GATS1Z

0.0169

Moreau–Broto autocorrelation of lag 1 weighted by atomic number

Autocorrelation

ATS5pe

0.0167

Moreau–Broto autocorrelation of lag 5 weighted by Pauling EN

TopoPSA

TopoPSA

0.0161

Topological polar surface area

Autocorrelation

ATS5se

0.0160

Moreau–Broto autocorrelation of lag 5 weighted by Sanderson EN

Autocorrelation

ATS4p

0.0150

Moreau–Broto autocorrelation of lag 4 weighted by polarizability

Autocorrelation

ATS4i

0.0144

Moreau–Broto autocorrelation of lag 4 weighted by ionization potential

Autocorrelation

GATS1are

0.0131

Geary coefficient of lag 1 weighted by Allred–Rocow EN

Autocorrelation

GATS1pe

0.0126

Geary coefficient of lag 1 weighted by Pauling EN

MoeType

VSA_EState3

0.0122

VSA EState Descriptor 3 (5.00 ≤ x < 5.41)

Autocorrelation

ATS5are

0.0122

Moreau–Broto autocorrelation of lag 5 weighted by Allred–Rocow EN

Autocorrelation

GATS1m

0.0122

Geary coefficient of lag 1 weighted by mass

Autocorrelation

ATSC1are

0.0115

Centered Moreau–Broto autocorrelation of lag 1 weighted by Allred–Rocow EN

Autocorrelation

ATSC1c

0.0112

Centered Moreau–Broto autocorrelation of lag 1 weighted by Gasteiger charge

InformationContent

BIC0

0.0111

0-ordered bonding information content

Autocorrelation

GATS2Z

0.0110

Geary coefficient of lag 2 weighted by atomic number

AtomCount

nAtom

0.0109

Molecular ID on O atoms