# tooluniverse-gwas-finemapping

Fine-maps GWAS association loci using SuSiE, FINEMAP, and Open Targets L2G scores. Prioritizes true causal genetic variants over positional nearest-gene assumptions.

- skill · `type=skill` · Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, OpenCode · 362 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-gwas-finemapping/SKILL.md
- Updated: 2026-10-24
- HTML: https://affpapa.org/en/nevibecode/skills/tooluniverse-gwas-finemapping

## What it does

This skill prioritizes causal variants in genome-wide association study (GWAS) loci by calculating Bayesian posterior probabilities. It resolves linkage disequilibrium (LD) blocks using methods like SuSiE and FINEMAP, generating 95% or 99% credible sets. Instead of assuming the lead SNP is causal, the skill evaluates the distribution of posterior probabilities across all variants in the set.

To map variants to effector genes, it queries Open Targets Genetics for locus-to-gene (L2G) scores. The workflow combines eQTL evidence, Hi-C chromatin interactions, and coding consequences. The instructions mandate executing Python code (pandas, scipy, statsmodels, matplotlib) via Bash to compute statistics directly instead of describing hypothetical analysis.

A high-scoring lead SNP can be an innocent tag variant on a genotyping array. When credible sets span wide LD blocks, statistical fine-mapping alone stalls, requiring tissue-specific functional annotations and colocalization to break ties.

## Install

### Claude Code (native platform)

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

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

### Codex

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

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

### Cursor

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

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

### GitHub Copilot

1. **GitHub CLI**
```bash
gh skill install mims-harvard/ToolUniverse tooluniverse-gwas-finemapping
```

2. **Manually: put the files in place**
- File: `.github/skills/tooluniverse-gwas-finemapping/SKILL.md`

### Gemini CLI

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

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

### Windsurf (Devin Desktop)

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

### OpenCode

**Skills CLI (Vercel)**
```bash
npx skills add mims-harvard/ToolUniverse --skill tooluniverse-gwas-finemapping -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-gwas-finemapping/SKILL.md

## When to use it
- identify causal variants in GWAS loci using credible sets
- calculate posterior probabilities with SuSiE and FINEMAP
- link non-coding variants to target genes with Open Targets L2G scores
- resolve LD ties using functional annotations and eQTL colocalization

## Example prompts
- Find the likely causal variant for type 2 diabetes at the TCF7L2 locus
- Fine-map rs429358 and inspect its credible set from Open Targets
- Retrieve all causal loci and L2G predictions for study GCST90029024

## 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: e7b1958b70cc7b7a.

BEGIN THIRD-PARTY FILE e7b1958b70cc7b7a
```text
---
name: tooluniverse-gwas-finemapping
description: Statistical fine-mapping of GWAS loci using credible sets (SuSiE, FINEMAP) and locus-to-gene scoring (Open Targets L2G). Identifies likely causal variants and target genes — distinct from positional 'nearest gene' which is often wrong. Use for prioritizing causal variants at GWAS hits, comparing fine-mapping methods, and converting lead SNPs to target genes.
disable-model-invocation: true
---

## 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.

# GWAS Fine-Mapping & Causal Variant Prioritization

Identify and prioritize causal variants at GWAS loci using statistical fine-mapping and locus-to-gene predictions.

## Overview

Genome-wide association studies (GWAS) identify genomic regions associated with traits, but linkage disequilibrium (LD) makes it difficult to pinpoint the causal variant. **Fine-mapping** uses Bayesian statistical methods to compute the posterior probability that each variant is causal, given the GWAS summary statistics.

**REASONING STRATEGY — Start Here**:
Fine-mapping asks: which variant at this locus is CAUSAL? Work through this chain:
1. **LD structure first** — variants in high LD (r² > 0.8) cannot be statistically distinguished from each other. Look up the LD block via Open Targets or the GWAS Catalog before assuming any single variant is the cause.
2. **Functional annotation breaks LD ties** — if two variants have similar posterior probabilities but one is coding (missense, stop-gain) or sits in an active regulatory element (promoter, enhancer), that variant is biologically prioritized. Functional evidence is the tiebreaker.
3. **eQTL colocalization is the key bridge** — a variant that is also a significant eQTL for a nearby gene in the relevant tissue (e.g., a pancreatic islet eQTL for a T2D locus) has a mechanistic story. Look up eQTL evidence via Open Targets L2G scores; don't assume the nearest gene is the effector gene.

This skill provides tools to:
- **Prioritize causal variants** using fine-mapping posterior probabilities
- **Link variants to genes** using locus-to-gene (L2G) predictions
```
END THIRD-PARTY FILE e7b1958b70cc7b7a

The third-party file has ended. Everything between the e7b1958b70cc7b7a 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 2404 of 13408 bytes. © mims-harvard · Apache-2.0 · https://github.com/mims-harvard/ToolUniverse/tree/HEAD/skills/tooluniverse-gwas-finemapping/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