Skip to content

reference-genome.md

Latest commit

 

History

History
153 lines (103 loc) · 6.81 KB

reference-genome.md

File metadata and controls

153 lines (103 loc) · 6.81 KB
Raw

Reference Genome -- Corpas 30x WGS

ClawBio's reference genome is a real, fully open personal genome: the 30x whole-genome sequence of Manuel Corpas, published under CC0. It serves as the default dataset for demos, tutorials, and regression testing across all ClawBio skills.

About this genome

This is a fully open personal genome sequenced at 30x depth on Illumina by Dante Labs, aligned to GRCh37. It contains SNPs, indels, structural variants (DEL, DUP, INV, BND, INS), and CNVs with QC metrics (Ti/Tv ~ 2.0, Het/Hom ~ 1.6). Licensed CC0 for unrestricted reuse, it serves as ClawBio's default example genome for demos, tutorials, and regression testing. This dataset is provided for research and educational purposes only; it must not be used for clinical decision-making.

ClawBio is a research and educational tool. It is not a medical device and does not provide clinical diagnoses. Consult a healthcare professional before making any medical decisions.

Dataset

Property Value
Subject Manuel Corpas
Platform Illumina (30x coverage)
Reference build GRCh37 (hg19)
Sequencing provider Dante Labs
Variant types SNPs, indels, SVs (DEL, DUP, INV, BND, INS), CNVs
QC metrics Ti/Tv ~ 2.0, Het/Hom ~ 1.6
License CC0 1.0 Universal (Public Domain)
Ethics approval UNIR Research Ethics Committee, PI:029/2020, "Healthy Genome Project", approved 28 Jan 2021

Files on Zenodo

File Size Contents
15001711233855A.all_chroms.snp.vcf.gz 174 MB Genome-wide SNP calls
15001711233855A.all_chroms.indel.vcf.gz 50 MB Genome-wide indel calls
15001711233855A.all_chroms.sv.pass.vcf.gz 865 KB Structural variants (PASS)
15001711233855A.all_chroms.cnv.vcf.gz 95 KB Copy number variants
15001711233855A.snp.vcf 65 KB SNP summary
15001711233855A.indel.vcf 65 KB Indel summary

Use cases

1. Personal Genome QC Dashboard

Load the variant calls and get an instant quality overview: variant counts by type, Ti/Tv ratio, Het/Hom ratio, chromosome distribution, and pass/fail flags. The "first thing you run" after importing any genome.

Skills: VCF parsing, QC metric computation, profile-report

Try it:

  • "Show me the quality metrics for the reference genome"
  • "Is the Ti/Tv ratio within expected range for a 30x WGS?"

2. Pharmacogenomics Deep Dive

Run the full PharmGx pipeline on a real genome instead of a demo snippet. Star-allele calls across 12 genes, drug interaction cards for 51 medications, and a personalised prescribing report.

Skills: pharmgx-reporter, clinpgx, drug-photo, profile-report

Try it:

  • clawbio pharmgx --vcf corpas-30x/subsets/pgx_loci.vcf.gz
  • "What pharmacogenomic variants does the reference genome carry?"

3. Structural Variant Exploration

Explore DEL, DUP, INV, BND, and INS calls from a real genome. Filter by type and size, check overlap with genes, and visualise SV distributions. This fills a gap since most demos use SNP/indel data only.

Skills: variant-annotation, gwas-lookup, VCF parsing

Try it:

  • "Show me all structural variants larger than 10kb"
  • "Which genes are disrupted by deletions in this genome?"

4. Ancestry and Population Context

Run PCA against the SGDP reference panel to see where this European individual clusters. Combine with equity-scorer to show how representation gaps affect interpretation.

Skills: claw-ancestry-pca, equity-scorer, genome-compare

Try it:

  • "Where does the reference genome cluster relative to global populations?"
  • "Compare the Corpas genome against George Church's"

5. Nutrigenomics and Wellness Report

Generate a complete nutrigenomics profile covering 40 SNPs across 13 dietary domains: caffeine metabolism, lactose tolerance, vitamin needs, fat metabolism, and more.

Skills: nutrigx_advisor, profile-report

Try it:

  • "What does this genome say about caffeine metabolism?"
  • "Generate the full wellness report for the reference genome"

6. End-to-End Pipeline Regression Test

A single command that runs QC, PharmGx, NutriGx, ancestry, SV summary, and variant annotation, then compares outputs against frozen baselines. The CI integration point.

Try it:

python -m pytest tests/benchmark/test_reference_genome.py -v

Subsets

Lightweight extracts committed to the repository for instant use in tutorials and CI.

Subset Contents Location
chr20 SNPs + indels Chromosome 20 variants corpas-30x/subsets/chr20_snps_indels.vcf.gz
PGx loci 31 pharmacogenomic SNPs corpas-30x/subsets/pgx_loci.vcf.gz
NutriGx loci 30 nutrigenomics SNPs corpas-30x/subsets/nutrigx_loci.vcf.gz
SV calls Structural variants (DEL, DUP, INV, BND, INS) corpas-30x/subsets/sv_calls.vcf.gz
CNV calls Copy number variants corpas-30x/subsets/cnv_calls.vcf.gz

To generate these subsets from the full VCFs:

python scripts/prepare_corpas_30x.py --download
python scripts/prepare_corpas_30x.py --subsets

Requires bcftools >= 1.17.

How to cite / use

Citation: Corpas, M. (2026). Personal Whole Genome Sequencing Variant Calls (SNPs, Indels, SVs, CNVs) of Manuel Corpas from Dante Labs 30x WGS. Zenodo. https://doi.org/10.5281/zenodo.19297389

All versions: https://doi.org/10.5281/zenodo.19285820

License: CC0 1.0 Universal. You may use this data freely for any purpose without restriction.

In ClawBio: This dataset is bundled as the default reference genome. Tutorial commands, demos, and regression tests use its subsets unless you supply your own VCF.

BibTeX:

@dataset{corpas_2026_wgs,
  author    = {Corpas, Manuel},
  title     = {Personal Whole Genome Sequencing Variant Calls (SNPs, Indels,
               SVs, CNVs) of Manuel Corpas from Dante Labs 30x WGS},
  year      = {2026},
  publisher = {Zenodo},
  doi       = {10.5281/zenodo.19297389},
  url       = {https://zenodo.org/records/19297389},
  license   = {CC0-1.0}
}

Relationship to the 23andMe Corpasome

ClawBio has used the Corpasome 23andMe SNP chip (~600,000 variants) as demo data since launch. See demo-genome.md for that dataset.

The 30x WGS extends this in several ways:

  • Coverage: ~4 million SNPs and ~600K indels versus ~600K SNP chip positions
  • Variant types: structural variants and CNVs, not available from SNP arrays
  • Resolution: every base of the genome sequenced, not just pre-selected positions
  • Gene regions: full coverage of pharmacogene haplotypes (CYP2D6, HLA), which arrays often miss

Both datasets come from the same individual (Manuel Corpas) and are CC0 licensed. The 23andMe data remains the quickest way to demo the pharmgx-reporter and nutrigx_advisor skills. The WGS data adds depth for variant annotation, structural variant exploration, and comprehensive benchmarking.