Dataset Card for LucaVirus-OpenVirus-Gene

1. Dataset Summary

LucaVirus-OpenVirus-Gene is a large-scale genomic dataset consisting exclusively of viral nucleotide sequences. It is a specialized subset of the OpenVirus corpus, curated specifically for the pre-training of the LucaVirus-Gene foundation model.

By focusing purely on viral genomes, this dataset provides a high-density corpus of 10.4 million sequences, enabling models to capture the intricate evolutionary patterns, regulatory motifs, and genomic architectures of DNA and RNA viruses.

2. Dataset Statistics

The dataset focuses solely on nucleotide sequences (genomes, genes, and fragments):

Feature	Count / Description
Total Sequences	10.4 Million
Sequence Type	Nucleotide (DNA/RNA)
obj_type Identifier	gene (Exclusive)
Primary Use	Pre-training for LucaVirus-Gene

3. Data Structure & Format

3.1 File Organization

The dataset is provided as a compressed .tar archive. Upon extraction, the data is partitioned into three standard machine-learning subsets:

LucaVirus-OpenVirus-Gene/dataset/v1.0/
├── train/          # Training set (primary corpus for genomic pre-training)
├── dev/            # Validation set (for model selection and tuning)
└── test/           # Test set (for final evaluation and benchmarking)

Each directory (train, dev, test) contains one or more CSV files with headers.

3.2 CSV Schema

All CSV files follow a consistent four-column schema:

Column Name	Description	Details
obj_id	Sample ID	Unique identifier for each viral sequence.
obj_type	Sequence Type	Set to gene for all entries in this dataset (Nucleotide).
obj_seq	Sequence Content	Raw nucleotide string (A, T(U), C, G, N).
obj_label	Label	Metadata, taxonomic info, or functional labels associated with the genome (Annotation, Bio Knowledge).

4. Intended Use

• Genomic Foundation Modeling: Building models like LucaVirus-Gene that specialize in the "language of genomes."
• Viral Evolution Studies: Analyzing conserved nucleotide patterns across divergent viral lineages.
• Regulatory Element Discovery: Identifying viral gene boundaries, promoters, and other non-coding functional motifs.

5. Usage Example

You can extract the archive and load the genomic data using the following Python snippet:

import tarfile
import pandas as pd
import os

# 1. Extract the genomic dataset
with tarfile.open("LucaVirus-OpenVirus-Gene.tar.gz", "r:gz") as tar:
    tar.extractall(path="./LucaVirus-OpenVirus-Gene")

with tarfile.open("LucaVirus-OpenVirus-Gene/dataset.tar.gz", "r:gz") as tar:
    tar.extractall(path="./LucaVirus-OpenVirus-Gene/dataset")
    
# 2. Load a sample from the training set
train_path = "./LucaVirus-OpenVirus-Gene/dataset/v1.0/train"
csv_files = [f for f in os.listdir(train_path) if f.endswith('.csv')]

if csv_files:
    # Load the first CSV file
    df = pd.read_csv(os.path.join(train_path, csv_files[0]))
    
    # Verify the sequence type
    print(f"Loaded {len(df)} genomic sequences.")
    print(df[['obj_id', 'obj_seq']].head())

6. Related Resources

This dataset is a core component of the LucaGroup biological modeling ecosystem. - Full Corpus (Gene + Prot): LucaVirus-OpenVirus-Gene-Prot - Protein Subset: LucaVirus-OpenVirus-Prot - Models: Visit the LucaVirus Collection.

7. Citation

If you use this dataset in your research, please cite:

@article{lucavirus2025,
  title={Predicting the Evolutionary and Functional Landscapes of Viruses with a Unified Nucleotide-Protein Language Model: LucaVirus.},
  author={Pan, Yuan-Fei* and He, Yong*. et al.},
  journal={bioRxiv},
  year={2025},
  url={https://www.biorxiv.org/content/early/2025/06/20/2025.06.14.659722}
}

8. License

This dataset is released under the MIT License.

9. Contact

For further information, please visit the LucaGroup GitHub, email to: [YongHe: sanyuan.hy@alibaba-inc.com, heyongcsat@gmail.com], or contact the team via the Hugging Face organization profile.