An Attention Graph Neural Network for Stereo-active Molecules

Introduction

Molecules can show stereochemistry: two molecules with the same atomic connectivity may exhibit different bioactivity due to different spatial arrangements. We propose a graph neural network architecture that utilizes a chiral-sensitive aggregation function and self-attention mechanism to improve the performance of molecular properties prediction by exploiting chiral information. Unlike many black-box deep learning models, the internals of our network are interpretable by visualizing the learned weights of the attention layers, providing better support for drug discovery.

Requirement

• python >= 3.5
• torch >= 1.7
• torchvision >= 0.8
• torchaudio >= 0.7

Installation

PyTorch version:

pip install torch==1.7.1+cu110 torchvision==0.8.2+cu110 torchaudio===0.7.2 -f https://download.pytorch.org/whl/torch_stable.html

Install PyTorch Geometric:

pip install -q torch-scatter -f https://pytorch-geometric.com/whl/torch-1.9.0+cu102.html
pip install -q torch-sparse -f https://pytorch-geometric.com/whl/torch-1.9.0+cu102.html
pip install -q torch-cluster -f https://pytorch-geometric.com/whl/torch-1.9.0+cu102.html
pip install -q torch-geometric

Usage

To run the optimal model, run the following command:

git clone https://github.com/sangttruong/stereonet.git
cd stereonet
pip install -r requirements.txt
bash run/optimal_exp.sh

To test:

bash run/optimal_test.sh

To do residual analysis:

bash run/optimal_resid_diag.sh

Acknowledgement

This project was started in 12/2020 by Sang Truong and Quang Tran under the mentorship of Professor Brian Howard. We thank Lucky Pattanaik for his support in understanding the theory of permutation-equivariant aggregation function. We thank Professor Jeffrey Hansen, Professor Bridget Gourley, Professor Suman Balasubramanian for their support on the chemical and mathematical foundation of the project, as well as allowing Sang Truong to complete this project as a part of his interdisciplinary major in Computational Chemistry at DePauw University.

Citation

@inproceedings{
    stereonet,
    title={An Attention Graph Neural Network for Stereo-active Molecules},
    author={Sang Truong and Quang Tran},
    booktitle={CMD-IT/ACM Richard Tapia Celebration of Diversity in Computing Conference},
    year={2021}
}