---
license: unknown
task_categories:
- tabular-classification
- graph-ml
- text-classification
tags:
- chemistry
- biology
- medical
pretty_name: MoleculeNet BBBP
size_categories:
- 1K<n<10K
configs:
- config_name: default
  data_files:
  - split: train
    path: "bbbp.csv"
---

# MoleculeNet BBBP

BBBP (Blood-Brain Barrier Penetration) dataset [[1]](#1), part of MoleculeNet [[2]](#2) benchmark. It is intended to be used through 
[scikit-fingerprints](https://github.com/scikit-fingerprints/scikit-fingerprints) library.

The task is to predict blood-brain barrier penetration (barrier permeability) of small drug-like molecules.

| **Characteristic** | **Description** |
|:------------------:|:---------------:|
|        Tasks       |        1        |
|      Task type     |  classification |
|    Total samples   |       2039      |
|  Recommended split |     scaffold    |
| Recommended metric |      AUROC      |

## References
<a id="1">[1]</a> 
Ines Filipa Martins et al.
"A Bayesian Approach to in Silico Blood-Brain Barrier Penetration Modeling"
J. Chem. Inf. Model. 2012, 52, 6, 1686–1697
https://pubs.acs.org/doi/10.1021/ci300124c

<a id="2">[2]</a> 
Wu, Zhenqin, et al.
"MoleculeNet: a benchmark for molecular machine learning."
Chemical Science 9.2 (2018): 513-530
https://pubs.rsc.org/en/content/articlelanding/2018/sc/c7sc02664a