powellgenomicslab/scancestry-reference-data

scAncestry Reference Panel

Reference data for scancestry, a tool for inferring genetic ancestry from single-cell genomics data.

Reference genome build: GRCh38 / hg38.

Contents

This dataset bundles imputation, phasing, and population-reference files used by the scAncestry pipeline:

• gnomad.genomes.v3.1.2.hgdp_tgp.miss0.01.maf0.01.vcf.gz (+ .tbi) — gnomAD v3.1.2 HGDP+1000G common variants, used for PCA reference
• gnomad.genomes.v3.1.2.hgdp_tgp.miss0.01.maf0.01.geno.gz — same variants in LASER/TRACE .geno format (decompressed client-side at download)
• gnomad.genomes.v3.1.2.hgdp_tgp.miss0.01.maf0.01.pruned.RefPC.coord — pre-computed reference PC coordinates for TRACE
• gnomad.genomes.v3.1.2.hgdp_tgp.phased_rare_snps.ancestry.miss0.05_updated_ids_updated_freqs_af01.frq — population allele frequencies for fastNGSadmix
• gnomad.genomes.v3.1.2.hgdp_tgp.phased_rare_snps.ancestry.miss0.05_updated_ids.nInd — per-population sample counts for fastNGSadmix
• ref_ancestries_dt.csv — sample-to-population mapping for plotting
• ref_vcfs_glimpse/ — per-chromosome phased reference VCFs for GLIMPSE imputation
• ref_vcfs_cellsnplite_chr/, ref_vcfs_cellsnplite_non_chr/ — TOPMed-derived candidate-SNP VCFs for cellsnp-lite (two builds: chr1-style and 1-style notation)
• glimpse_chunks/ — GLIMPSE imputation chunk definitions and per-chromosome genetic maps
• chr_vcf_locations_*.csv, glimpse_chunks.csv — manifest CSVs consumed by the pipeline (path placeholders are rewritten client-side at download)
• chr_names_mapping_chr_nonchr.txt — chromosome name mapping between UCSC and Ensembl/NCBI styles
• checksums.md5 — MD5 hashes for all files; validated client-side at download

Total size: ~185 GiB on this repo, expanding to ~273 GiB on disk after the bundled .geno.gz is decompressed.

Usage

The intended way to use this dataset is via the scAncestry tool's downloader script, which also runs the post-download validation, decompression, and path-rewriting steps:

SIF=/path/to/scancestry_0.3.0.sif
singularity exec "$SIF" /scancestry/scripts/download_ref.sh /path/to/ref_data

To download the raw files directly without post-processing:

hf download powellgenomicslab/scancestry-reference-data \
    --repo-type dataset --local-dir /path/to/ref_data

If you take this path you'll need to do the post-processing yourself (or via singularity exec "$SIF" /scancestry/scripts/configure_ref.sh /path/to/ref_data).

Provenance & licensing

The data here is derived from several public human-genetics resources. Each source has its own terms that propagate through this dataset; by using this data you agree to comply with the upstream licenses.

• gnomAD v3.1.2 — Broad Institute. Public release; see gnomAD terms.
• 1000 Genomes Project — International Genome Sample Resource. Public-use data; see IGSR data reuse policy.
• HGDP (Human Genome Diversity Project) — Stanford / CEPH. Public release through gnomAD.
• TOPMed — used here for candidate-SNP coordinates only (no genotype data). See TOPMed data use terms.
• GLIMPSE genetic maps — bundled per-chromosome from the GLIMPSE project.

Citation

If you use this dataset, please cite the scAncestry tool and the underlying resources:

@software{scancestry,
  title = {scAncestry: genetic ancestry inference from single-cell genomics data},
  url   = {https://github.com/powellgenomicslab/scancestry},
  year  = {2026}
}

@article{karczewski2020gnomad,
  title   = {The mutational constraint spectrum quantified from variation in 141,456 humans},
  author  = {Karczewski, Konrad J. and others},
  journal = {Nature},
  volume  = {581},
  pages   = {434--443},
  year    = {2020},
  doi     = {10.1038/s41586-020-2308-7}
}

@article{1000genomes2015global,
  title   = {A global reference for human genetic variation},
  author  = {{1000 Genomes Project Consortium}},
  journal = {Nature},
  volume  = {526},
  pages   = {68--74},
  year    = {2015},
  doi     = {10.1038/nature15393}
}

@article{bergstrom2020hgdp,
  title   = {Insights into human genetic variation and population history from 929 diverse genomes},
  author  = {Bergstr{\"o}m, Anders and others},
  journal = {Science},
  volume  = {367},
  pages   = {eaay5012},
  year    = {2020},
  doi     = {10.1126/science.aay5012}
}

@article{rubinacci2021glimpse,
  title   = {Efficient phasing and imputation of low-coverage sequencing data using large reference panels},
  author  = {Rubinacci, Simone and others},
  journal = {Nature Genetics},
  volume  = {53},
  pages   = {120--126},
  year    = {2021},
  doi     = {10.1038/s41588-020-00756-0}
}

Versioning

This dataset is a snapshot bundled with scAncestry release 0.3.0. Future updates will be tagged on the repository.