# mol_gen

**Repository Path**: z9527567/mol_gen

## Basic Information

- **Project Name**: mol_gen
- **Description**: No description available
- **Primary Language**: Unknown
- **License**: GPL-3.0
- **Default Branch**: main
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2025-10-04
- **Last Updated**: 2025-10-04

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# Using GNN property predictors as molecule generators (DIDgen)

This is the repository for the paper: [*Using GNN property predictors as molecule generators*](https://doi.org/10.1038/s41467-025-59439-1).

You can use DIDgen (**D**irect **I**nverse **D**esign **gen**erator) to generate diverse molecules with a specific property by *inverting* a GNN that predicts that property.

<img src="https://github.com/ftherrien/inv-design/blob/master/anim89.gif" width="200" height="200"><img src="https://github.com/ftherrien/inv-design/blob/master/anim32.gif" width="200" height="200"><img src="https://github.com/ftherrien/inv-design/blob/master/anim111.gif" width="200" height="200">

## Install

```
pip install git+https://github.com/ftherrien/inv-design
```

## Usage

If you have trouble using or installing DIDgen please [create an issue](https://github.com/ftherrien/inv-design/issues/new) or [ask a question](https://github.com/ftherrien/inv-design/discussions/new?category=q-a). I will be happy to help! 

### As a command-line interface (cli)

```
didgenerate [-h] [-n N] [-c CONFIG] [-o OUTDIR]
```

The results are organized in `OUTDIR` as such

```
OUTDIR
├── drawings
│   └── generated_mol_0.png     # An image(s) of the generated graph(s)
├── final_performance_data.pkl
├── final_performance.png
├── initial_mol.png
├── model_weights.pth
├── property_value_list.txt     # A list of smiles strings and corresponding predicted property
├── qm9/
└── xyzs/
    ├── generated_mol_0.pickle  # A RDKit mol object
    └── generated_mol_0.xyz     # A molecular conformer with 3D positions
```

### As a Python API

```python
from didgen import generate

out = generate(number_of_samples, outdir, config_dict)
```

This creates the same output directory as the cli. `out` is a list of python dictionaries containing the generated graphs, their corresponding smiles and the predicted property.

### Parameters

You can find a list of parameters and their description in [the documentation](https://github.com/ftherrien/inv-design/blob/master/docs/parameters.md).

## Generate molecules online using Colab

[Train a GNN to predict the energy gap on a subset of QM9 and generate a molecule with an energy gap of 4.1 eV](https://colab.research.google.com/github/ftherrien/inv-design/blob/master/didgenerate.ipynb)

## Citation

```
@Article{Therrien2025,
author={Therrien, F{\'e}lix and Sargent, Edward H. and Voznyy, Oleksandr},
title={Using GNN property predictors as molecule generators},
journal={Nature Communications},
year={2025},
month={May},
day={08},
volume={16},
number={1},
pages={4301},
issn={2041-1723},
doi={10.1038/s41467-025-59439-1},
url={https://doi.org/10.1038/s41467-025-59439-1}
}
```