gocam-curationlisted
Install: claude install-skill ai4curation/ai-gene-review
# GO-CAM curation & review
Use this skill to **read and assess** GO-CAM models. This project is read-only:
we never edit or upload models. We cache them (`gocams/<id>/<id>-src.yaml`),
index their activities (`gocams/index.tsv`), and record assessments in
`gocams/<id>/<id>-review.yaml` (schema class `GoCamReview`).
## What an activity (annoton) is
Each GO-CAM activity couples:
- `enabled_by` — the **gene product** carrying the activity;
- `molecular_function` — the GO MF term;
- `part_of` — optional **biological process** it contributes to;
- `occurs_in` — optional **cellular component** it happens in;
- `causal_associations` — directed edges to **downstream** activities.
In the cached YAML these are fields of each entry under `activities:`; they are
flattened one-row-per-activity into `gocams/index.tsv` (join key:
`gene_product`).
## How to evaluate an activity
For each activity, work through these and record the outcome in the
`GoCamActivityReview`:
1. **Molecular function** — most specific correct term; not a bare `binding`
term; the real role (catalytic / receptor / adaptor / sequestering).
→ `references/molecular-function.md`
2. **`has input`** — names the substrate / target-gene / effector / cargo, not a
ligand or raw DNA. → `references/has-input.md`
3. **Causal edges** — direct vs indirect, subject→object directionality,
mechanism vs phenotype. → `references/causal-relations.md`
4. **Complexes** — right active subunit vs complex term.
→ `reference