dna-seq on GitHubJust-DNATry the demo

Install, then ask

A module is a table of claims, each tied to its paper. just-module-creator helps your assistant write, check and compile one. It needs uv and Python 3.13.

Claude Code

Terminal
claude plugin marketplace add dna-seq/dna-seq-claude-marketplace
claude plugin install just-module-creator@dna-seq

Codex

Terminal
codex plugin marketplace add dna-seq/dna-seq-claude-marketplace
codex plugin add just-module-creator@dna-seq
Then, in the assistant
/create-module caffeine metabolism, from this paper

How a module gets made

The steps

  1. Open a folder with the tables the module needs.
  2. Draft from a source such as ClinVar, CPIC, ClinPGx or CIViC.
  3. Decide what survives what each genotype means, and which paper backs it.
  4. Place each variant on the genome, noting where each position came from.
  5. Check what was written against what the sources say.
  6. Compile into Parquet with a fingerprint, so it rebuilds the same.
  7. Publish when you say so.

The habits

A lookup shows a value, it does not paste it
A later check compares the row with the same source.
A check that could not run stays unknown
Nothing turns a failed lookup into a pass.
A quote is a sentence found in the paper
The assistant fetches the article and quotes it word for word.
Most candidate rows get dropped
In the FTO example, six of seven were, and its README says why.

Publishing

Rehearse on the polygon, where test versions can be deleted. A version on the catalog stays, so a module that installed yesterday installs the same way tomorrow.

Before publishing, the plugin can run your draft on genomes on your own machine through Just-DNA-Lite. Drug response, star alleles and repeat counts have table kinds of their own: what more a module can carry.