ConfGAN

Requirements

• python (version>=3.8.0)
• tensorflow-gpu (version>=2.6.0)
• rdkit (version>=2023.03.1)
• numpy (version==1.20.3)
• networkx (version>=3.1)

Installation

Create a conda environment named ConfGAN from env.yml :

conda env create --file env.yml

Data

Download the GEOM raw dataset: here.

Process raw data using the command: python ./data/process_data.py --dataset QM9/Drugs

The processed data can be downloaded from here: Google Drive .

Usage

Generate Conformations

Generating conformations is very simple and can be done using three methods.

The first method is using mol files.

python main.py --generate --mol test.mol --num_conf 50

The second method is to input SMILES

python main.py --generate --smiles 'CNC(=O)c1nc(C)cs1' --num_conf 50

The third method is to use a CSV file for batch generation, where the CSV file format: smiles, num_conf 'CNC(=O)c1nc(C)cs1', 50 'CNC(=O)c1nc(N)cs1', 20

python main.py --generate True --csv test.csv

More generation options can be found in generate.py.

Train

Example: training a model for QM9 molecules.

python main.py --train --train_dataset ./data/qm9_train.pkl --val_dataset ./data/qm9_val.pkl

More training options can be found in main.py.

Citation

Contact

12049045@mail.sustech.edu.cn

Updates

• Oct. 19, 2024.