GenPept-Curated-2025

Attribution: This is a mirror/re-upload of the GenPept-Curated-2025 benchmark dataset by Huynh Trong Pham, Bao Huynh, and Thanh-Hoang Nguyen-Vo (bioRxiv 2026). All data and credit belong to the original authors. Distributed here under the same CC-BY-NC 4.0 license with attribution; this upload is a convenience mirror only and is not affiliated with the authors.

• Original source: https://github.com/biochem-data-sci/GenPept-Curated-2025
• Paper (bioRxiv): https://doi.org/10.64898/2026.04.25.720793
• License: CC-BY-NC 4.0 (non-commercial)

Description

A curated, class-balanced benchmark for antimicrobial peptide (AMP) prediction with homology-controlled partitioning.

• 11,000 peptide sequences (5,500 AMP / 5,500 non-AMP), fully balanced.
• Derived from Bacteria, Archaea, and Fungi GenPept/NCBI Protein records.
• Sequence lengths: 10–200 amino acids.
• Each sequence uses only the 20 standard amino acids.
• Constructed via a reproducible pipeline: taxonomic scoping → quality control → precursor handling → annotation-based labeling → IPG-based deduplication.

Files

File	Description
dataset.csv	Standard downstream input table (11k rows). Key cols: accession_version, sequence, length, label, length_bin, tax_id, taxon_name, description, cds_products, cds_genes, ipg_id, refseq_accession, precursor_flag.
balanced_11000.csv	Compact convenience table (11k rows). Cols: label (AMP/non-AMP), sequence, length.

Usage

Load with the datasets library (auto-parses the CSV):

from datasets import load_dataset

ds = load_dataset("devansh0703/GenPept-Curated-2025")
print(ds["train"].num_rows)          # 11000
print(ds["train"][0])                # {'label': 'AMP', 'sequence': ..., 'length': 40}

Data Columns (dataset.csv)

• accession_version — GenPept accession.version
• sequence — amino acid sequence (10–200 aa, 20 standard amino acids)
• length — sequence length (int)
• label — AMP or non-AMP (binary classification target)
• length_bin — length bucket (e.g. 10-50)
• tax_id / taxon_name — NCBI taxonomy (Bacteria / Archaea / Fungi)
• description — original GenPept definition
• cds_products / cds_genes — CDS annotations
• ipg_id — NCBI Identical Protein Groups cluster ID
• refseq_accession — RefSeq accession when available
• precursor_flag — whether the record is a precursor

Note on splits

The official homology-controlled CD-HIT split (train/val/test at 90% identity) is generated by the upstream 03_cluster_split_cdhit.py script — it is not bundled here as static files. This dataset ships the raw curated tables. See the upstream repo for the split generation.

Citation

If you use this dataset, please cite the original authors:

@article{Pham2026GenPeptCurated2025,
  title={GenPept-Curated-2025: A Benchmark Dataset for Antimicrobial Peptide Prediction with Homology-Controlled Partitioning},
  author={Huynh Trong Pham and Bao Huynh and Thanh-Hoang Nguyen-Vo},
  journal={bioRxiv},
  year={2026},
  doi={10.64898/2026.04.25.720793}
}

License

CC-BY-NC 4.0 International (non-commercial). See https://creativecommons.org/licenses/by-nc/4.0/