scientific-statgen-playbook

scientific-statgen-playbook

A scientific-software-playbook-compatible plugin collection for downstream statistical genetics: GWAS analysis, fine-mapping, heritability, TWAS, and biological validation against public databases.

How It Fits Together

Layer	Repository	Focus
Upstream	scientific-software-playbook	Model design, inference code, structured dev workflow
Downstream	This repo	GWAS analysis, fine-mapping, heritability, TWAS
Validation	This repo	Biological validation via OpenTargets, GWAS Catalog, Ensembl

Install both to get an end-to-end agent-assisted statistical genetics workflow.

Plugins

statgen-analysis

GWAS analysis workflows: SuSiE fine-mapping, LDSC heritability, TWAS simulation, FUSION TWAS, and publication-ready visualization.

statgen-validation

Biological validation of findings against public databases:

• OpenTargets — gene-disease associations, target tractability, genetic constraint
• GWAS Catalog — known associations by variant, gene, or trait
• Ensembl VEP — variant functional annotation (consequence, SIFT, PolyPhen, CADD)

Installation

Claude Code

# Install as a plugin
git clone https://github.com/CamelliaRui/scientific-statgen-playbook.git ~/.claude/plugins/scientific-statgen-playbook

# Or project-level
git clone https://github.com/CamelliaRui/scientific-statgen-playbook.git .claude/plugins/scientific-statgen-playbook

OpenAI Codex

git clone https://github.com/CamelliaRui/scientific-statgen-playbook.git .skills/scientific-statgen-playbook

Other Agent Skills-compatible tools

Clone the repo and point your tool to the directory. The plugins follow the Agent Skills open standard and the scientific-software-playbook plugin format.

Install Dependencies

R (required for SuSiE and FUSION):

install.packages(c("susieR", "data.table", "jsonlite", "optparse", "glmnet"))

Python (required for LDSC, TWAS, visualization, validation):

uv pip install pandas numpy matplotlib plotly openpyxl seaborn scipy scikit-learn requests

Quick Start

Once installed, just ask in natural language:

"Estimate the SNP heritability for my height GWAS using EUR reference"
"Run SuSiE on my GWAS summary stats with the provided LD matrix"
"Validate my top fine-mapping variants against GWAS Catalog and OpenTargets"
"Annotate these credible set SNPs with VEP consequences"
"Check if PCSK9 has known associations with LDL cholesterol in OpenTargets"

What's Included

Plugin	Tool	What it does
analysis	SuSiE	Fine-map causal variants from GWAS summary stats or individual data
analysis	LDSC	SNP heritability, genetic correlations, partitioned heritability
analysis	TWAS Simulator	Simulate TWAS for power analysis and methods development
analysis	FUSION TWAS	Gene-trait associations using GTEx v8 weights (49 tissues)
analysis	Visualization	Publication-ready plots (locus zoom, Manhattan, PIP, heatmaps)
validation	OpenTargets	Gene-disease evidence, tractability, genetic constraint
validation	GWAS Catalog	Known associations by variant, gene, or trait
validation	Ensembl VEP	Variant annotation (consequence, SIFT, PolyPhen, CADD)

Project Structure

scientific-statgen-playbook/
├── .claude-plugin/
│   └── marketplace.json            # Plugin marketplace registry
├── plugins/
│   ├── statgen-analysis/           # GWAS analysis plugin
│   │   ├── .claude-plugin/
│   │   │   └── plugin.json
│   │   ├── skills/statgen-analysis/SKILL.md
│   │   ├── scripts/               # SuSiE, LDSC, TWAS, FUSION
│   │   ├── reference/             # Detailed docs
│   │   ├── visualization/         # Publication-ready plots
│   │   ├── examples/              # Example data and tutorials
│   │   └── tests/
│   └── statgen-validation/         # Biological validation plugin
│       ├── .claude-plugin/
│       │   └── plugin.json
│       ├── skills/biological-validation/SKILL.md
│       ├── scripts/               # OpenTargets, GWAS Catalog, Ensembl
│       ├── reference/             # API docs
│       └── tests/
├── AGENTS.md                       # Plugin asset source of truth
├── README.md
└── pytest.ini

Using with the Scientific Software Playbook

These plugins are designed to complement the scientific-software-playbook. When both are installed: