/CIGIN

AAAI 2020: Chemically Interpretable Graph Interaction Network for Prediction of Pharmacokinetic Properties of Drug-like Molecules

Primary LanguageJupyter NotebookMIT LicenseMIT

CIGIN : Chemically Interpretable Graph Interaciton Network

Official implementation of CIGIN presented at proceedings of the 34th AAAI conference on Artificial Intelligence, AAAI-20.

CIGIN is a chemically interpretable graph interaction network for prediction of pharmacokinetic properties of drug-like molecules.

Requirements:

  • PyTorch
  • Numpy
  • RDKit

Usage:

  • Examples for prediction and analysis of interaction between solute and solvent atoms are given in the notebook.
  • Required scripts are given in the scripts folder.
  • Trained model weight is provided in weights folder.

People

  • Yashaswi Pathak
  • Siddhartha Laghuvarapu
  • Sarvesh Mehta
  • U. Deva Priyakumar

If you find this useful in your research, please cite:

@inproceedings{pathak2020chemically,
title={Chemically interpretable graph interaction network for prediction of pharmacokinetic properties of drug-like 		  molecules},
  author={Pathak, Yashaswi and Laghuvarapu, Siddhartha and Mehta, Sarvesh and Priyakumar, U Deva},
  booktitle={Proceedings of the AAAI Conference on Artificial Intelligence},
  volume={34},
  number={01},
  pages={873--880},
  year={2020}
}