A small project that implements residual gated graph convolutional networks https://arxiv.org/pdf/1711.07553v2.pdf in pytorch and predicts ADMET properties of molecules Comes with a UI built in gradio
Uses a multi-headed network approach as using a singular backbone allows for feature sharing and possibly better results.
python main.py train
python main.py evaulate
python ui.py
TDC Dataset: https://github.com/mims-harvard/TDC
@article{Huang2021tdc,
title={Therapeutics Data Commons: Machine Learning Datasets and Tasks for Drug Discovery and Development},
author={Huang, Kexin and Fu, Tianfan and Gao, Wenhao and Zhao, Yue and Roohani, Yusuf and Leskovec, Jure and Coley,
Connor W and Xiao, Cao and Sun, Jimeng and Zitnik, Marinka},
journal={Proceedings of Neural Information Processing Systems, NeurIPS Datasets and Benchmarks},
year={2021}
}
Implicit valence = atom valence - valence from bond connections