Learning Multimodal Graph-to-Graph Translation for Molecular Optimization

Abstract

We view molecular optimization as a graph-to-graph translation problem. The goal is to learn to map from one molecular graph to another with better properties based on an available corpus of paired molecules. Since molecules can be optimized in different ways, there are multiple viable translations for each input graph. A key challenge is therefore to model diverse translation outputs. Our primary contributions include a junction tree encoder-decoder for learning diverse graph translations along with a novel adversarial training method for aligning distributions of molecules. Diverse output distributions in our model are explicitly realized by low-dimensional latent vectors that modulate the translation process. We evaluate our model on multiple molecular optimization tasks and show that our model outperforms previous state-of-the-art baselines.

References in corpus (6)

• Semi-Supervised Classification with Graph Convolutional Networks
• Deep Convolutional Networks on Graph-Structured Data
• Hierarchical Multiscale Recurrent Neural Networks
• Grammar Variational Autoencoder
• Covariant Compositional Networks For Learning Graphs
• Generating Focussed Molecule Libraries for Drug Discovery with Recurrent Neural Networks

Cited by in corpus (7)

• Deep Learning and Knowledge-Based Methods for Computer Aided Molecular Design -- Toward a Unified Approach: State-of-the-Art and Future Directions
• Generalization and Representational Limits of Graph Neural Networks
• Learning to Make Generalizable and Diverse Predictions for Retrosynthesis
• Hierarchical Inter-Message Passing for Learning on Molecular Graphs
• Generalizable Resource Allocation in Stream Processing via Deep Reinforcement Learning
• Graph Polish: A Novel Graph Generation Paradigm for Molecular Optimization
• CORE: Automatic Molecule Optimization Using Copy & Refine Strategy