/MolHGCN

A Hypergraph Convolutional Neural Network for Molecular Properties Prediction using Functional Group

Primary LanguagePython

MolHGCN

A Hypergraph Convolutional Neural Network for Molecular Properties Prediction using Functional Group [paper]

Requirements

dgl 0.5.3
dgllife 0.2.6
networkx 2.5
torch 1.7.0
scikit-learn 0.22.1
RDKit

Dataset

Datasets can be found in: https://bit.ly/34wuwJo

Quickstart

  • molhgcn_classification.py and molhgcn_regression.py are for the classification and regression tasks of MolHGCN respectively.
  • ablation_classification_atom.py and ablation_regression_atom.py are for the MolHGCN-AtomGC experiments in Section 4.2.1 respectively.
  • ablation_classification_fg.py and ablation_regression_fg.py are for the MolHGCN-FuncGC experiments in Section 4.2.1 respectively.