Skip to main content

Table 3 Description of the properties being predicted for each dataset along with dataset sizes and property values ranges

From: Extended study on atomic featurization in graph neural networks for molecular property prediction

Dataset name

Predicted property

Range

Number of samples

Human

Metabolic stability \(\hbox {T}_{1/2}\) (log h)

0, 4.8

3578

Rat

Metabolic stability \(\hbox {T}_{1/2}\) (log h)

0, 4.5

1819

QM9

Atomization free energy at room temperature (eV)

− 714.6, − 117.8

134K

ESOL

Aqueous solubility (log M/L)

− 4.1, 2.2

1128

  1. The property values were not normalized for the model training