huggingworld/ncbi-refseq-complete

Dataset Card for NCBI RefSeq Complete

Dataset Summary

The NCBI Reference Sequence (RefSeq) complete dataset provides a comprehensive, integrated, non-redundant, well-annotated set of sequences, including genomic DNA, transcripts, and proteins. It serves as a stable reference for genome annotation, gene identification and characterization, mutation and polymorphism analysis, expression studies, and comparative analyses.

This dataset has been processed into a parquet format suitable for machine learning training, consisting of genomic sequences and their corresponding accessions.

Supported Tasks and Leaderboards

This dataset can be used to train genomic foundational models (e.g., DNA language models), perform sequence classification, or predict functional regions.

Languages

The data is genomic sequence data (DNA/RNA) represented as strings of nucleotides (A, C, G, T, N).

Dataset Structure

Data Instances

Each instance in the dataset represents a continuous sequence of nucleotides.

{
  "accession": "NC_000001.11_part1",
  "sequence": "GATCGATCGATC..."
}

Data Fields

• accession: The NCBI access identifier for the sequence (e.g., NM_001301717.1). Large sequences may be partitioned with _partXXX suffixes to limit chunk sizes.
• sequence: The raw nucleotide sequence strings.

Data Splits

The dataset is provided as a single train split containing all sequences.

Dataset Creation

Curation Rationale

The RefSeq database is curated by NCBI to provide a non-redundant set of standards for genomic sequences.

Source Data

• Initial Data Collection and Normalization: The data is compiled by NCBI from various sources including GenBank, and then heavily curated and annotated.
• Who are the source language producers? The data represents genetic material from various organisms, gathered by scientists globally and curated by NCBI.

Considerations for Using the Data

Social Impact of Dataset

Genomic datasets can have a profound impact on medical research, drug discovery, and biological understanding.

Discussion of Biases

As with any genomic database, species that are heavily researched (e.g., humans, model organisms, pathogens) are significantly overrepresented compared to the general biosphere.

Additional Information

Dataset Curators

The database is curated by the National Center for Biotechnology Information (NCBI).

Licensing Information

Data from NCBI RefSeq is in the public domain and freely available for use.

ML Dataset Attribution

This dataset (RefSeq Release 235) was retrieved and optimally processed by huggingworld. The sequence data has been specifically serialized into parquet format using the latest pyarrow standards, targeted specifically for ML Research and massive-scale model training. If you utilize this processed version of the dataset in your research or applications, please cite or acknowledge this repository.

Citation Information

@misc{huggingworld2026ncbirefseq,
  author       = {huggingworld},
  title        = {NCBI RefSeq Complete (Release 234) March 20, 2026 - Processed for ML Research},
  year         = {2026},
  publisher    = {Hugging Face},
  url          = {https://ftp.ncbi.nlm.nih.gov/refseq/release}
  howpublished = {\url{https://huggingface.co/datasets/huggingworld/ncbi-refseq-complete}}
}

@article{10.1093/nar/gkv1189,
    author = {O'Leary, Nuala A and Wright, Mathew W and Brister, J Rodney and Ciufo, Stacy and Haddad, Diana and McVeigh, Rich and Rajput, Bhanu and Robbertse, Barbara and Smith-White, Brian and Ako-Adjei, Danso and Astashyn, Alexander and Badretdin, Azat and Bao, Yiming and Blinkova, Olga and Brover, Vyacheslav and Chetvernin, Vyacheslav and Choi, Jinna and Cox, Eric and Ermolaeva, Olga and Farrell, Christopher M and Goldfarb, Tamara and Hajrullaj, Tatijana and Hoffman, Stephen and Joardar, Vinita and Katz, Kenneth S and Kelly, Christopher and Kemmer, Douglas and Komatsoulis, George and Klimke, William and Lang, Gang and McCabe, Vanessa and Murphy, Terence and Peluso, Paul and Pujar, Shashikant and Richards, Stephen and Schriml, Lynn and Seaberry, Carl and Sequeira, Eneida and Sherbert, Craig and Slotta, Douglas and Thibaud-Nissen, Francoise and Tomashevsky, Mikhail and Vatsan, Raman and Wallin, Craig and Webb, David and Wilmer, Thomas and Wu, Wendy and Zuiderwijk, Anne and Pruitt, Kim D},
    title = "{Reference sequence (RefSeq) database at NCBI: current status, taxonomic expansion, and functional annotation}",
    journal = {Nucleic Acids Research},
    volume = {44},
    number = {D1},
    pages = {D733-D745},
    year = {2015},
    month = {11},
    issn = {0305-1048},
    doi = {10.1093/nar/gkv1189},
    url = {https://doi.org/10.1093/nar/gkv1189},
    eprint = {https://huggingface.co/datasets/huggingworld/ncbi-refseq-complete/resolve/main/gkv1189.pdf},
}