JT-VAE for the Dual-Objective Inverse Design of Metal Complexes

This repo contains the modified JT-VAE code for the publication "A Deep Generative Model for the Inverse Design of Transition Metal Ligands and Complexes"

Requirements

The environment.yml file is an export of a conda environment that can run this model.

Important: The version of RDKit is very important. For newer versions of RDKit the model does not work! The tree decomposition will give kekulization errors with newer versions of RDKit.

Code for model training

• fast_molvae/ contains codes for unconditional JT-VAE training. Please refer to fast_molvae/README.md for details.
• fast_jtnn/ contains codes for model and data implementation.
• fast_molopt/ contains codes for training a conditional JT-VAE and for performing conditional optimization with a trained model.
• data/ contains various ligand training data.

FastJTNNpy3

The code is based on a fork of FastJTNNpy3.