metadata
license: mit
task_categories:
- tabular-regression
- graph-ml
- text-classification
tags:
- chemistry
- biology
- medical
pretty_name: Biogen ADME Solubility
size_categories:
- 1K<n<10K
configs:
- config_name: default
data_files:
- split: train
path: solubility.csv
Biogen ADME Solubility
Biogen_ADME_Solubility dataset from the Biogen ADME benchmark [1]. It is intended to be used through scikit-fingerprints library.
The task is to predict log10 of aqueous solubility at pH 6.8 (in ug/mL) of molecules.
| Characteristic | Description |
|---|---|
| Tasks | 1 |
| Task type | regression |
| Total samples | 2173 |
| Recommended split | time |
| Recommended metric | MAE |
References
[1] Fang, Cheng, et al. "Prospective Validation of Machine Learning Algorithms for Absorption, Distribution, Metabolism, and Excretion Prediction: An Industrial Perspective" J. Chem. Inf. Model. 2023, 63, 11, 3263-3274 https://doi.org/10.1021/acs.jcim.3c00160