MOLRGEN : A Training and Evaluation Setting for De Novo Molecular Generation with Reasoning Models

Welcome to MolRGen, a comprehensive framework for molecular generation tasks with integrated protein-ligand docking evaluation. This project provides datasets, benchmarks, and a reward server for training and evaluating models that generate drug-like molecules optimized for specific biological targets.

Quick Start

Installation

Basic Installation:

git clone https://github.com/Fransou/MolRGen.git
cd MolRGen
pip install -e .[main]

If called as pip install ., pytdc will not be installed.

Note: This installation requires OpenBabel to be installed on your system. OpenBabel is used for molecular file format conversions and processing.

Running the Reward Server

export DOCKING_ORACLE=autodock_gpu
... # Set other environment variables as needed
export DATA_PATH=... # Path to your data directory
uvicorn --host 0.0.0.0 --port 8000 molrgen.server:app

Using the API

import requests

response = requests.post(
    "http://localhost:8000/get_reward",
    json={
        "query": "CC(C)Cc1ccc(cc1)C(C)C(=O)O",
        "prompt": "Generate a drug-like molecule...",
        "metadata": [
            {
                "properties": ["QED", "protein_1"],
                "objectives": ["above", "minimize"],
                "target": [0.7, 0.0]
            }
        ]
    }
)

⚙️ Reward Server API We use AutoDock-GPU for fast GPU-accelerated docking calculations. The Molecular Verifier server is built using FastAPI, and supports concurrent requests, ensuring efficient handling of multiple docking evaluations, and asynchroneous pipelines.

🤖 MCP Server Support The server also includes MCP (Model Context Protocol) support for seamless integration with AI workflows and language models that use MCP for tool calling and function execution.

Citation

If you use MolRGen in your research, please cite:

...

License

Apache License 2.0. See LICENSE for details.

Support

For issues, questions, or contributions, please visit our GitHub repository.