just-dna-seq/ensembl_variations

Ensembl Variations (Parquet Format)

This dataset contains Ensembl human genetic variations converted to Parquet format for fast and efficient VCF annotation.

Dataset Structure

The dataset is organized by variant type:

• SNV/
• deletion/
• indel/
• insertion/
• sequence_alteration/
• substitution/

Each directory contains parquet files split by chromosome.

Why Parquet?

Parquet format provides significant advantages over VCF for annotation tasks:

• 10-100x faster querying and filtering
• Efficient compression (smaller file sizes)
• Column-based access (read only what you need)
• Native support in modern data processing tools (Polars, DuckDB, Arrow)
• Schema evolution support

Usage

With Polars (Recommended)

import polars as pl

# Load SNV variants for chromosome 21
df = pl.scan_parquet("hf://datasets/just-dna-seq/ensembl_variations/data/SNV/homo_sapiens-chr21.parquet")

# Filter variants by position
variants = df.filter(
    (pl.col("POS") >= 10000000) & (pl.col("POS") <= 20000000)
).collect()

print(variants)

With DuckDB

import duckdb

# Query variants directly from HuggingFace
result = duckdb.sql("""
    SELECT * FROM 'hf://datasets/just-dna-seq/ensembl_variations/data/SNV/homo_sapiens-chr21.parquet'
    WHERE POS BETWEEN 10000000 AND 20000000
    LIMIT 10
""").df()

print(result)

For VCF Annotation

from prepare_annotations.annotation.logic import VCFAnnotator

# Annotate your VCF file with Ensembl data
annotator = VCFAnnotator("your_variants.vcf")
annotated_df = annotator.annotate_with_ensembl(
    ensembl_path="hf://datasets/just-dna-seq/ensembl_variations/data/"
)

Schema

Typical columns in the parquet files:

• CHROM: Chromosome (e.g., "chr1", "chr2", ...)
• POS: Position (1-based)
• ID: Variant ID (rsID)
• REF: Reference allele
• ALT: Alternate allele(s)
• QUAL: Quality score
• FILTER: Filter status
• INFO_*: Various INFO field columns
• TSA: Variant type annotation (SNV, deletion, indel, insertion, substitution, sequence_alteration)

Citation

If you use this dataset, please cite:

@misc{ensembl_variations_parquet,
  title = {Ensembl Variations (Parquet Format)},
  author = {GenoBear Team},
  year = {2026},
  howpublished = {\url{https://huggingface.co/datasets/just-dna-seq/ensembl_variations}},
  note = {Processed from Ensembl Variation Database}
}

And the original Ensembl data:

@article{martin2023ensembl,
  title={Ensembl 2023},
  author={Martin, Fergal J and others},
  journal={Nucleic acids research},
  volume={51},
  number={D1},
  pages={D933--D941},
  year={2023},
  publisher={Oxford University Press}
}

License

This dataset is released under Apache 2.0 license. The original Ensembl data is available under their terms of use.

Maintenance

This dataset is maintained by the GenoBear project. For issues or questions:

• GitHub: https://github.com/dna-seq/just-dna-lite
• HuggingFace: https://huggingface.co/just-dna-seq