chip-atlas

name	chip-atlas
aliases	["chip atlas","atlas-dna","atlas-genetics","/chipatlas","/atlas_report"]
description	Standalone Atlas/23andMe raw DNA analyzer that generates a Russian HTML health plan and can optionally download large GWAS/ClinVar references for a deeper local report. No secrets, no external upload of DNA data.
metadata	{"clawdbot":{"emoji":"🧬","command":"/chipatlas"}}

chip-atlas

Generate a human-readable HTML genetic health plan from an Atlas.ru / 23andMe-style raw genotype file.

The skill has two levels:

1. Fast curated report — portable, no downloads, embedded marker panel.
2. Full-reference report — explicitly offers the user to download large public GWAS/ClinVar references, explains why, gives concrete links, and then uses downloaded files to enrich the report.

Operator behavior

When a user asks for a serious/complete report, do not silently stay in lightweight mode. Explain:

• default mode is quick and useful;
• full-reference mode downloads large public files;
• these files are needed to scan the raw DNA against broad GWAS/ClinVar references;
• the user's DNA stays local;
• this is educational, not medical advice.

Then offer the full workflow:

python3 bin/chip_atlas.py refs-info
python3 bin/chip_atlas.py download-refs --all --dir data
python3 bin/chip_atlas.py analyze path/to/raw_dna.txt \
  --refs-dir data \
  --out output/full_reference_report.html \
  --md output/full_reference_report.md

Use --yes only for non-interactive runs where the user already approved downloading large files:

python3 bin/chip_atlas.py download-refs --all --dir data --yes

Input format

Expected TSV-like raw genotype file:

# rsid	chromosome	position	genotype
rs9939609	16	53820527	AA
rs4988235	2	136608646	AG

The parser accepts tab or whitespace separated columns and ignores comment lines starting with #.

CLI

Fast report:

python3 bin/chip_atlas.py analyze path/to/atlas_raw.txt --out output/report.html

Markdown too:

python3 bin/chip_atlas.py analyze path/to/atlas_raw.txt --out output/report.html --md output/report.md

Stats:

python3 bin/chip_atlas.py stats path/to/atlas_raw.txt

Reference explanation and concrete links:

python3 bin/chip_atlas.py refs-info

Download references:

python3 bin/chip_atlas.py download-refs --all --dir data

Expanded report after download:

python3 bin/chip_atlas.py analyze path/to/atlas_raw.txt \
  --refs-dir data \
  --out output/full_reference_report.html \
  --md output/full_reference_report.md

Reference sources

• GWAS Catalog associations TSV
  https://www.ebi.ac.uk/gwas/api/search/downloads/alternative
  Use: broad research associations between rsID markers and published traits.

• ClinVar GRCh38 VCF
  https://ftp.ncbi.nlm.nih.gov/pub/clinvar/vcf_GRCh38/clinvar.vcf.gz
  Use: clinical-reference submissions that require doctor/lab confirmation.

These downloads can be hundreds of MB and may become >1GB as references grow. They live in data/ and are ignored by git.

Output style

The generated HTML follows the practical “human report” style:

• no rsID-heavy section titles;
• plain Russian explanations;
• colored labels: опасно, осторожно, можно, уточнить;
• every section says what it means and what to do;
• optional “Расширенные публичные базы” section when refs are provided;
• action checklist at the end;
• strong medical disclaimer.

Done criteria

A good complete run should:

1. parse the raw file;
2. show SNP count and curated-marker count;
3. generate fast standalone HTML/Markdown;
4. offer full-reference download with concrete URLs and reason;
5. after download, generate expanded HTML/Markdown with GWAS/ClinVar matches;
6. avoid uploading DNA data;
7. keep medical claims cautious.

Medical disclaimer

This is not medical advice, not diagnosis, and not a clinical genetic test. Consumer raw DNA files can have strand/build/provider quirks. Any medication, pregnancy, oncology, thrombosis, or treatment decision must be confirmed with a physician and a certified lab.