Learning Topology-Specific Experts for Molecular Property Prediction

This is Official Pytorch Implementation for the paper "Learning Topology-Specific Experts for Molecular Property Prediction". Suyeon Kim, Dongha Lee, SeongKu Kang, Seonghyeon Lee, Hwanjo Yu (AAAI-23)

The paper is available at Link.

Run

python main.py --dataset bbbp

We refer the baseline code to build our implementation. https://github.com/snap-stanford/pretrain-gnns

Package Install

conda create -n topexpert python=3.8

conda activate topexpert

conda install pytorch==1.8.0 torchvision==0.9.0 torchaudio==0.8.0 cudatoolkit=11.1 -c pytorch -c conda-forge

conda install -c conda-forge rdkit

conda install pytorch-geometric -c rusty1s -c conda-forge

Cite (Bibtex)

• If you find TopExpert useful in your research, please consider citing:

@article{kim2023learning,
  title={Learning Topology-Specific Experts for Molecular Property Prediction},
  author={Suyeon Kim, Dongha Lee, SeongKu Kang, Seonghyeon Lee, Hwanjo Yu},
  booktitle={AAAI},
  year={2023}
}