jiahuizhang/PepBenchData

This dataset is part of PepBenchmark, a standardized benchmark for peptide machine learning introduced in the paper PepBenchmark: A Standardized Benchmark for Peptide Machine Learning.

PepBenchmark unifies datasets, preprocessing, and evaluation protocols for peptide drug discovery. It comprises three components:

• PepBenchData: A well-curated collection of 29 canonical-peptide and 6 non-canonical-peptide datasets across 7 groups.

• PepBenchPipeline: A standardized preprocessing pipeline ensuring consistent data cleaning, construction, and splitting.

• PepBenchLeaderboard: A unified evaluation protocol and leaderboard with strong baselines across major methodological families.

• GitHub Repository: https://github.com/ZGCI-AI4S-Pep/PepBenchmark

• Paper: https://huggingface.co/papers/2604.10531

Sample Usage

You can use the pepbenchmark library to load and manage the datasets:

from pepbenchmark.dataset_manager.single_dataset import SinglePeptideDatasetManager

# Initialize the manager for a specific dataset
manager = SinglePeptideDatasetManager(
    "ace_inhibitory",
    official_feature_names=["fasta", "label"],
    dataset_dir="../PepBenchData/PepBenchData-50",
)

# Access features
sequences = manager.get_feature("fasta")
labels = manager.get_feature("label")

# Set data splits
splits = manager.set_official_split_indices(
    split_type="hybrid_split",
    fold_seed=0
)

print(f"Train samples: {len(splits['train'])}")
print(f"Validation samples: {len(splits['valid'])}")
print(f"Test samples: {len(splits['test'])}")

Citation

@inproceedings{zhang2026pepbenchmark,
  title={PepBenchmark: A Standardized Benchmark for Peptide Machine Learning},
  author={Zhang, Jiahui and Wang, Rouyi and Zhou, Kuangqi and Xiao, Tianshu and Zhu, Lingyan and Min, Yaosen and Wang, Yang},
  booktitle={International Conference on Learning Representations (ICLR)},
  year={2026},
  url={https://openreview.net/forum?id=NskQgtSdll}
}