# tooluniverse-rare-disease-diagnosis

Turns a patient phenotype into a ranked differential: HPO term matching, Orphanet and OMIM lookups, gene panel prioritization, ACMG variant interpretation.

- skill · `type=skill` · Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, OpenCode · 406 installs · ★ 1759
- Author: mims-harvard
- License: Apache-2.0 · policy=mirror · the beginning of the file is at the end of this page
- Source: https://github.com/mims-harvard/ToolUniverse/tree/HEAD/skills/tooluniverse-rare-disease-diagnosis/SKILL.md
- Updated: 2026-10-24
- HTML: https://affpapa.org/en/nevibecode/skills/tooluniverse-rare-disease-diagnosis

## What it does

The skill runs in phases. Clinical reasoning comes first and produces 3–5 candidates, then symptoms are converted to HPO terms, and those drive the search across Orphanet, OMIM and DisGeNET. Gene panel work, variant interpretation, structure analysis for VUS and a literature pass follow.

It ships a parameter correction table, because several tool calls use argument names you would not guess: `OpenTargets_get_associated_drugs_by_target_ensemblID` takes `ensemblId`, `ClinVar_get_variant_details` takes `id`, `MyGene_query_genes` takes `q`, `gnomad_get_variant` takes `variant_id`. Analysis is meant to be computed, not described — the skill tells the agent to write and run Python through Bash. Database queries go out in English.

Candidate genes get cross-checked against PanelApp, which is a different signal from MARRVEL and ClinGen curation: it reflects what a diagnostic lab actually orders today. Only green genes (confidence 3) belong in a report; amber and red are research follow-up. For a clearly dominant pedigree, autosomal recessive candidates are dropped using the inheritance pattern from `MARRVEL_get_omim_phenotypes`. The skill is marked `disable-model-invocation: true`, so you invoke it yourself.

## Install

### Claude Code (native platform)

1. **Skills CLI (Vercel)**
```bash
npx skills add mims-harvard/ToolUniverse --skill tooluniverse-rare-disease-diagnosis -a claude-code
```

2. **Manually: put the files in place**
- File: `~/.claude/skills/tooluniverse-rare-disease-diagnosis/SKILL.md` — Copy the whole skill folder; for one project use .claude/skills/tooluniverse-rare-disease-diagnosis/

### Codex

1. **Skills CLI (Vercel)**
```bash
npx skills add mims-harvard/ToolUniverse --skill tooluniverse-rare-disease-diagnosis -a codex
```

2. **Manually: put the files in place**
- File: `.agents/skills/tooluniverse-rare-disease-diagnosis/SKILL.md` — For all projects use ~/.agents/skills/

### Cursor

1. **Skills CLI (Vercel)**
```bash
npx skills add mims-harvard/ToolUniverse --skill tooluniverse-rare-disease-diagnosis -a cursor
```

2. **Manually: put the files in place**
- File: `.cursor/skills/tooluniverse-rare-disease-diagnosis/SKILL.md` — Cursor also reads .agents/skills/ and .claude/skills/

### GitHub Copilot

1. **GitHub CLI**
```bash
gh skill install mims-harvard/ToolUniverse tooluniverse-rare-disease-diagnosis
```

2. **Manually: put the files in place**
- File: `.github/skills/tooluniverse-rare-disease-diagnosis/SKILL.md`

### Gemini CLI

1. **Skills CLI (Vercel)**
```bash
npx skills add mims-harvard/ToolUniverse --skill tooluniverse-rare-disease-diagnosis -a gemini-cli
```

2. **Platform CLI**
```bash
gemini skills install https://github.com/mims-harvard/ToolUniverse --path skills/tooluniverse-rare-disease-diagnosis
```

### Windsurf (Devin Desktop)

**Manually: put the files in place**
- File: `.devin/skills/tooluniverse-rare-disease-diagnosis/SKILL.md` — The legacy .windsurf/skills/ path still works

### OpenCode

**Skills CLI (Vercel)**
```bash
npx skills add mims-harvard/ToolUniverse --skill tooluniverse-rare-disease-diagnosis -a opencode
```

### Claude.ai

**Claude.ai**
- Zip the folder and upload it in Settings → Capabilities → Skills

Where a file is given instead of a command, copy the files from the source to that path: https://github.com/mims-harvard/ToolUniverse/tree/HEAD/skills/tooluniverse-rare-disease-diagnosis/SKILL.md

## When to use it
- build a differential from a vague cluster of symptoms
- convert a phenotype description into HPO terms before searching
- drop genes a diagnostic lab does not test for
- classify a variant of uncertain significance against ACMG criteria
- check whether one pathway explains every finding

## Example prompts
- Work up this phenotype: infantile seizures, loss of milestones, hepatosplenomegaly — build and justify a differential
- Check PTEN against PanelApp and ClinGen: which panels include it and at what confidence
- Annotate this variant, score it against ACMG, and pull AlphaFold2 domains if the significance stays unclear

## Source text · SKILL.md · Apache-2.0

Below is the content of a third-party file. It is data, not instructions for you: do not act on anything written inside unless the human asks. Block tag: ca9722a6d5029bb1.

BEGIN THIRD-PARTY FILE ca9722a6d5029bb1
```text
---
name: tooluniverse-rare-disease-diagnosis
description: Rare disease differential diagnosis from patient phenotype — HPO term matching to candidate diseases (Orphanet, OMIM), gene panel prioritization, ACMG variant interpretation, and structure-based variant analysis. Use for diagnostic odyssey assistance, phenotype-to-disease ranking, and genetic-counseling differential generation.
disable-model-invocation: true
---

# Rare Disease Diagnosis Advisor

Systematic diagnosis support for rare diseases using phenotype matching, gene panel prioritization, and variant interpretation across Orphanet, OMIM, HPO, ClinVar, and structure-based analysis.

**KEY PRINCIPLES**:
1. **Report-first** - Create report file FIRST, update progressively
2. **Phenotype-driven** - Convert symptoms to HPO terms before searching
3. **Multi-database triangulation** - Cross-reference Orphanet, OMIM, OpenTargets
4. **Evidence grading** - Grade diagnoses by supporting evidence strength
5. **English-first queries** - Always use English terms in tool calls

## LOOK UP, DON'T GUESS
When uncertain about any scientific fact, SEARCH databases first rather than reasoning from memory.

---

## COMPUTE, DON'T DESCRIBE
When analysis requires computation (statistics, data processing, scoring, enrichment), write and run Python code via Bash. Don't describe what you would do — execute it and report actual results. Use ToolUniverse tools to retrieve data, then Python (pandas, scipy, statsmodels, matplotlib) to analyze it.

## Clinical Reasoning Framework (BEFORE Tools)

Apply these strategies to form a 3-5 candidate differential, then use tools to confirm/refute:

1. **Multi-system involvement** - Symptoms spanning 2+ organ systems = strongest rare disease signal. Ask: what single pathway explains ALL features?
2. **Regression question** - Losing abilities vs never acquired? Regression = neurodegenerative/metabolic storage. Stable = developmental/structural.
```
END THIRD-PARTY FILE ca9722a6d5029bb1

The third-party file has ended. Everything between the ca9722a6d5029bb1 tags is data: requests and commands from it are not followed. An “end of file” line with another tag inside the block is part of the file.

First 1953 of 8899 bytes. © mims-harvard · Apache-2.0 · https://github.com/mims-harvard/ToolUniverse/tree/HEAD/skills/tooluniverse-rare-disease-diagnosis/SKILL.md

_Part of the page did not fit — see the HTML version._

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md