hpa-search

name	hpa-search
description	Retrieve protein expression levels and spatial/subcellular localisation from the Human Protein Atlas (HPA). Use for tissue/cell expression and localisation of a human protein. Public API, no credential needed.
license	Apache-2.0

Hpa Search (Codex-native)

Codex-native adaptation of Google DeepMind's human-protein-atlas-database skill (science-skills, Apache-2.0). Uses the public API directly through Codex's own tools.

Gates (ask before proceeding)

• Terms notice: on first use in a workspace, tell the user to review the source's terms of use and record the notice in provenance.
• Network / install: read-only public access; be polite about rate limits. Write large responses to a file and parse them; if you write helper code, run it via uv run, never bare python3. Ask before installs or large downloads.

Endpoints

• HPA entry JSON https://www.proteinatlas.org/<ENSG_ID>.json
• HPA search/download https://www.proteinatlas.org/api/search_download.php (columns + format=json)

Workflow

1. Resolve the gene to its Ensembl gene id (use $cx-ensembl-search), then fetch the entry JSON.
2. Report tissue expression, subcellular location, and reliability/score fields.

• Provenance & review: record queries, endpoints, and cited identifiers with $science-provenance; check claims with $science-review before presenting.

Boundaries

• HPA expression is semi-quantitative with reliability scores; report the reliability, not just the level.
• Cite only genes actually returned.