Claude Skill MartBrowse skillsWhat is a Skill?
 
← Back to list

NCATS ARAX Biomedical Knowledge Graph Lookup

Runs tightly constrained one-hop and endpoint-pinned two-hop queries against the NCATS Translator ARAX API, returning typed relationships with full provenance.

Data & AnalyticsAdvanced★ 33,030⑂ 3,248AI score 8/10Last updated: Aug 9, 2026

What it does

This skill lets Claude drive a standard-library Python client that wraps the ARAX production API from NCATS Translator:

  • Entity normalization — turn free text (e.g. "primary myelofibrosis") into a canonical CURIE plus Biolink category and synonym preview for review before any graph query.
  • One-hop lookup — typed subject/object with at least one endpoint pinned, optional qualifiers such as object_direction_qualifier=decreased.
  • Endpoint-pinned two-hop lookup — exactly one typed, unpinned intermediate node, with configurable expansion order.
  • Selected-provider federation — explicitly named 2–5 knowledge providers only, never "all KPs".
  • Provenance inspection — rebuild a bounded summary offline from saved request.json/response.json to review edge bindings, publications, and primary/aggregator sources.

The surface is deliberately narrow: no inference, ranking, link prediction, Pathfinder, batching, or raw TRAPI passthrough, and a hard 50-result cap. Zero results are reported as "not returned under these constraints," and result order is never presented as a rank.

Who it's for

  • Bioinformatics, pharmacology, and translational-research users comfortable with Biolink Model and TRAPI.
  • Teams that need reproducible, auditable query artifacts rather than LLM speculation.
  • Not for clinical guidance, patient data, or confidential target hypotheses — the skill explicitly forbids these.

Examples

  1. Imatinib ↔ ABL1: pin CHEBI:31690 and NCBIGene:25 with biolink:affects plus "activity_or_abundance / decreased" qualifiers, then inspect the returned edge's primary sources and publications.
  2. Ivacaftor → gene → cystic fibrosis: pin CHEBI:66901 and MONDO:0009061, leave a biolink:Gene intermediate, and enumerate candidate mediating genes with right-first expansion.
  3. Provider coverage comparison: rerun the same one-hop query with --mode federated --kp infores:rtx-kg2 --kp infores:molepro to compare source coverage; partial runs exit 7 but keep their artifacts.

· · · Install guide · · ·

Install in the Claude app (no terminal)
  1. Download the ZIP with the button below.
  2. In Claude, open Settings → Capabilities and turn on 'Code execution and file creation'. (one time)
  3. Go to Customize → Skills → + → 'Upload a skill' and upload the ZIP.
↓ Download ZIP
Install in Claude Code

Let Claude do it — paste this into Claude Code

Install the skill I found on Claude Skill Mart.
Copy the skills/ncats-arax folder from the GitHub repo K-Dense-AI/scientific-agent-skills into my ~/.claude/skills/ncats-arax/.
When it's done, tell me in one line what this skill can do.

Install with a command instead

git clone https://github.com/K-Dense-AI/scientific-agent-skills.git && mkdir -p ~/.claude/skills && cp -r scientific-agent-skills/skills/ncats-arax ~/.claude/skills/

⚠ This is a third-party skill. Check the source repository before installing.

  1. Prerequisites: Python 3.10+ and outbound HTTPS access to arax.transltr.io. No API key required.
  2. Clone the repo: git clone https://github.com/K-Dense-AI/scientific-agent-skills.git
  3. Install the skill: mkdir -p ~/.claude/skills && cp -r scientific-agent-skills/skills/ncats-arax ~/.claude/skills/
  4. Restart Claude Code and confirm ncats-arax appears in your skill list.
  5. Run preflight: python ~/.claude/skills/ncats-arax/scripts/arax_client.py preflight to verify the service and a supported TRAPI version.
  6. Try it: ask Claude something like "use ARAX to run a one-hop lookup between imatinib and ABL1." Artifacts land in the new, empty --output-dir you specify (request.json, response.json, summary.json).
  7. Safety note: queries and caller metadata may be publicly visible — never submit patient information or unpublished/proprietary research content.
View source on GitHub ↗License: MIT