The code and database for Graph Neural Network (GNN) of organic cage shape persistency and the explainability of the GNN.

  • filtered_all_smi.csv contains information of the cages in this study.

  • GNN_Train/ holds the code to build and train the GNN models.

  • Integrated_Gradient/ holds the calculation of integrated gradients to explain the prediction of GNNs.

  • trained_models/ holds the models trained in this study.

  • database/ holds the code to process the database to graph fingerprints.

Detailed information can be found in the info.txt files under each folder.

The related paper has been published in Digital Discovery (2022) DOI: 10.1039/D1DD00039J