Metadata-Version: 2.4
Name: gwaslab-agent
Version: 0.0.4
Summary: GWASLab Agent
Author-email: Yunye <gwaslab@gmail.com>
License-Expression: GPL-3.0-only
Project-URL: Homepage, https://cloufield.github.io/gwaslab/
Project-URL: Github, https://github.com/Cloufield/gwaslab
Classifier: Programming Language :: Python :: 3
Classifier: Operating System :: OS Independent
Requires-Python: <3.13,>=3.12
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: fastapi>=0.120.1
Requires-Dist: gwaslab>=3.6.8
Requires-Dist: langchain>=1.0.3
Requires-Dist: langchain-core>=1.0.3
Requires-Dist: langchain-google-genai>=3.0.3
Requires-Dist: langchain-openai>=1.0.3
Requires-Dist: mcp[cli]>=1.13.1
Requires-Dist: openpyxl>=3.1.5
Requires-Dist: rich>=14.1.0
Dynamic: license-file

## GWASLab-Agent

**GWASLab-Agent** is an LLM-powered framework for automated GWAS summary statistics processing, quality control, transformation, and visualization.  
It extends the original **GWASLab** Python package with intelligent planning, multi-step workflow generation, and agent-driven execution.

GWASLab-Agent is designed to serve as an *autonomous GWAS assistant*, capable of interpreting user instructions, planning complex operations, managing file paths, and producing publication-ready summaries and figures.

---

## Installation

1. Create a new environment (recommended)
```
# Create a clean environment with Python 3.12
conda create -n gwaslab-agent python=3.12

# Activate it
conda activate gwaslab-agent
```

2. Install GWASLab and GWASLab-Agent
```bash
pip install gwaslab
pip install gwaslab_agent
```

---

## Design of GWASLab-Agent

### SmartSumstats Object

At the core of GWASLab-Agent is the **SmartSumstats** object — an LLM-enhanced wrapper around `gl.Sumstats`.  
It integrates five coordinated sub-agents:

- **Loader** — detects file formats, parses paths, handles chromosome patterns  
- **Planner** — constructs optimal multi-step workflows  
- **Worker** — executes tasks, QC steps, and visualizations  
- **PathManager** — manages input/output paths and reference resources  
- **Summarizer** — generates structured summaries and Methods-section text  

Together, these sub-agents enable fully automated GWAS workflows with minimal user input.

---


---

## Citation

(Coming soon — please cite GWASLab and GWASLab-Agent once the corresponding manuscripts or preprints are available.)
