The Future of Biology Is Agentic

Bioinformatics is moving from scripts you babysit to skills your agent calls. ClawBio packages reproducible bioinformatics workflows — pharmacogenomics, drug-photo lookup, ClinPGx queries — as agent-callable skills that run locally and never leak your data. In this session we'll install ClawBio, run a real pharmacogenomics report end-to-end on demo genetic data, and watch an agent interpret CYP2D6 metabolizer status against CPIC drug guidelines.

Core idea: a VCF (or 23andMe / AncestryDNA file) goes in, a publication-grade pharmacogenomics report comes out — calls star alleles, assigns metabolizer phenotypes, and lists CPIC-backed dosage recommendations for 51 drugs across 12 genes. All in under a second, on your laptop.

If you want to try it before the workshop, this is the whole demo:

# Option A — Claude Code plugin
/plugin marketplace add ClawBio/ClawBio
/plugin install clawbio

# Option B — local install
git clone https://github.com/ClawBio/ClawBio.git && cd ClawBio
pip install -r requirements.txt
python clawbio.py run pharmgx --demo

The --demo flag ships with synthetic genotypes so you can see the full report without bringing your own data. Bring a 23andMe or AncestryDNA file if you want to run it on yourself — your data stays on your laptop.

• Laptop with macOS, Windows, or Linux. Admin rights to install Python packages.
• Python 3.10+ (or Claude Code with the plugin marketplace enabled — either works).
• Optional: a 23andMe .txt, AncestryDNA .csv, or VCF file you'd like to analyse on yourself.

No prior bioinformatics experience needed. If you've never written a VCF parser, that's the point — the skill does it for you. See Setup if you haven't installed an agent yet.