Zhouzhimeng/DrugEvolve-datasets

DrugEvolve Datasets

This repository contains the publicly available datasets associated with selected tasks reported in the main text of DrugEvolve, an AI system for autonomous algorithm evolution in drug development.

At the current stage, this repository includes data for a subset of the tasks evaluated in the DrugEvolve study. Additional datasets used in the study will be released in subsequent updates.

Repository structure

The currently available datasets are organized by task:

DrugEvolve-datasets/
├── ADMET_properties_prediction/
├── Drug_side-effect_prediction/
├── Molecular_docking/
├── Protein_druggable_site_annotation/
└── Target-specific_peptide_binder_design/

Each directory contains the data and supporting files associated with the corresponding DrugEvolve task.

Currently available tasks

Drug development stage	Task
Target identification	Protein druggable site annotation
Drug discovery	Molecular docking
Drug discovery	Target-specific peptide binder design
Drug discovery	ADMET properties prediction
Clinical trial	Drug side-effect prediction

These datasets correspond to selected tasks presented in the main text of the DrugEvolve study.

Use with DrugEvolve

The datasets in this repository are intended to support reproduction and evaluation of selected DrugEvolve applications.

The DrugEvolve source code, baseline implementations, and autonomous algorithm evolution framework are available in the DrugEvolve GitHub repository.

To obtain this dataset repository, use:

git clone https://huggingface.co/datasets/Zhouzhimeng/DrugEvolve-datasets

Alternatively, the repository can be downloaded using the Hugging Face Hub tools.

Data provenance and licensing

This repository contains task-specific datasets and supporting resources derived from multiple sources.

The original datasets may have different authors, licenses, citation requirements, and usage conditions. Therefore, no single redistribution license should be assumed to apply uniformly to all files in this repository.

Users are responsible for reviewing and complying with the license and usage requirements of the original data sources.

When using a specific dataset, please cite both:

1. the DrugEvolve study; and
2. the original publication, database, or benchmark associated with that dataset, where applicable.

Citation

If you use the datasets provided in this repository, please cite the DrugEvolve study:

@article{Zhou2026.08.16.745117,
  author = {Zhou, Zhimeng and Nan, Yang and Mou, Minjie and Qian, Yuntao and Liu, Yixiu and Zuo, Zhengyu and Yang, Hao and Xu, Weixian and Li, Bo and Jiang, Wanghao and Ren, Yanlin and Liao, Yang and Wang, Yimeng and Li, Yinghong and Yang, Qingxia and Xi, Zhiheng and Mi, Tiantian and Sun, Huaicheng and Liu, Pengfei and Zhu, Feng},
  title = {An AI System for Autonomous Algorithm Evolution in Drug Development},
  journal = {bioRxiv},
  year = {2026},
  doi = {10.64898/2026.08.16.745117},
  url = {https://www.biorxiv.org/content/10.64898/2026.08.16.745117v1}
}

Please also cite the original publications or databases associated with the individual datasets used in your experiments.

Related resources

• DrugEvolve paper
• DrugEvolve source code

Updates

This repository will be updated as additional DrugEvolve datasets become available for public release.

For questions regarding the datasets, data organization, or potential issues with source attribution and redistribution, please open an issue in the DrugEvolve GitHub repository.