biochemical-identifier-crosswalklisted
Install: claude install-skill timothymeyers/discovery-tools
# biochemical-identifier-crosswalk
Reconcile a reaction/compound/protein/taxon across model-native,
Rhea, ChEBI, EC, UniProt, NCBI Taxonomy, BioCyc/CyanoCyc, SABIO-RK, BRENDA
ligand, KEGG, and MetaNetX namespaces into one
`identifier-crosswalk.schema.json`-conformant record — while deliberately
preserving every ambiguity that a naive "one canonical ID per entity"
crosswalk would silently destroy.
## Dependencies
This skill is **not** standard-library only. `gh skill install` and plugin
install do not install Python packages, so install these before first use:
```bash
python3 -m pip install -r requirements.txt # PyYAML, jsonschema
```
`scripts/build_crosswalk.py` needs `PyYAML` to read the declarative entity
configs; `scripts/validate_crosswalk.py` needs `jsonschema`.
`scripts/formula_parser.py` is standard-library only and works without either.
## When to use
Trigger when the user asks anything like:
- "Map this Rhea/ChEBI ID to UniProt / MetaNetX / KEGG"
- "Is this really the same reaction/compound as that one?"
- "Build an identifier crosswalk for these reactions"
- "Resolve which direction this reaction actually runs"
- "This reaction has isoenzymes / paralogs — keep them separate"
- "Check mass and charge balance for this reaction"
- "Is a locus tag the same thing as a reaction ID?" (no)
Do NOT trigger for:
- Retrieving raw kinetic parameters (`sabio-rk-kinetics`,
`brenda-enzyme-data`) or gene→pathway network context
(`biocyc-network-context`) — this skill