Datasets:
metadata
license: unknown
task_categories:
- tabular-classification
- graph-ml
- text-classification
tags:
- chemistry
- biology
- medical
pretty_name: MoleculeNet HIV
size_categories:
- 10K<n<100K
configs:
- config_name: default
data_files:
- split: train
path: hiv.csv
MoleculeNet HIV
HIV dataset [1], part of MoleculeNet [2] benchmark. It is intended to be used through scikit-fingerprints library.
The task is to predict ability of molecules to inhibit HIV replication.
Characteristic | Description |
---|---|
Tasks | 1 |
Task type | classification |
Total samples | 41127 |
Recommended split | scaffold |
Recommended metric | AUROC |
References
[1] AIDS Antiviral Screen Data https://wiki.nci.nih.gov/display/NCIDTPdata/AIDS+Antiviral+Screen+Data
[2] 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