A Graph VAE and Graph Transformer Approach to Generating Molecular Graphs
This work addresses molecular generation, a domain-specific problem in chemistry and drug discovery, with incremental improvements in scalability and interpretability.
The authors tackled the problem of generating molecular graphs by proposing a model combining a variational autoencoder and a transformer with novel graph-specific layers, resulting in strong performance on the QM9 dataset for generating valid, unique, and novel molecules.
We propose a combination of a variational autoencoder and a transformer based model which fully utilises graph convolutional and graph pooling layers to operate directly on graphs. The transformer model implements a novel node encoding layer, replacing the position encoding typically used in transformers, to create a transformer with no position information that operates on graphs, encoding adjacent node properties into the edge generation process. The proposed model builds on graph generative work operating on graphs with edge features, creating a model that offers improved scalability with the number of nodes in a graph. In addition, our model is capable of learning a disentangled, interpretable latent space that represents graph properties through a mapping between latent variables and graph properties. In experiments we chose a benchmark task of molecular generation, given the importance of both generated node and edge features. Using the QM9 dataset we demonstrate that our model performs strongly across the task of generating valid, unique and novel molecules. Finally, we demonstrate that the model is interpretable by generating molecules controlled by molecular properties, and we then analyse and visualise the learned latent representation.