The official implementation of Graph LDA : Graph latent diffusion-based molecular representation learning for enhanced generalization in molecular property prediction
- Python 3.10.13
- PyTorch 2.4.0
- Pytorch geometric 2.5.3
- networkx 3.1
- RDKit version 2024.3.5.0
Dataset and pre-trained models in our experiments are avaiable here.
As mentioned in our paper, we use PIG-VAE for molecular graph autoencoder architecture. The code for the PIG-VAE can be found in pigvae_all/pigvae.