metadata
language: []
pretty_name: AQSOL
config_name: aqsol
dataset_size: 597796
size_categories: 10K<n<100K
license: cc-by-nc-sa-4.0
license_link: https://creativecommons.org/licenses/by-nc-sa/4.0/
tags:
- graphs
- molecules
- chemistry
task_categories:
- graph-regression
homepage: https://pubs.acs.org/doi/10.1021/ci3003352
aqsol
AQSOL molecular graphs with categorical atom/bond encodings for aqueous solubility regression.
License: CC BY-NC-SA 4.0
Splits
| Split | Rows | File | Size (MB) |
|---|---|---|---|
| train | 7,836 | train.parquet |
0.40 |
| validation | 998 | val.parquet |
0.08 |
| test | 999 | test.parquet |
0.12 |
Features
- mol_id: int64 unique identifier per molecule
- x: list[int64], shape (num_nodes,), atomic type categorical index
- edge_index: int64[2, num_edges], COO adjacency
- edge_attr: list[int64], shape (num_edges,), bond type categorical index
- num_nodes: int64 number of atoms
- y: float32 graph-level target (LogS)
Citation
@article{sorkun2019aqsol,
title={AQSOL: A Complete Aqueous Solubility Dataset for Small Molecules},
author={Sorkun, Murat C and Khetan, Aditya and Er, Sanli},
journal={Journal of Cheminformatics},
volume={11},
number={1},
pages={1--16},
year={2019},
publisher={Springer}
}