Constrained Graph Variational Autoencoders for Molecule Design

Abstract

Graphs are ubiquitous data structures for representing interactions between entities. With an emphasis on the use of graphs to represent chemical molecules, we explore the task of learning to generate graphs that conform to a distribution observed in training data. We propose a variational autoencoder model in which both encoder and decoder are graph-structured. Our decoder assumes a sequential ordering of graph extension steps and we discuss and analyze design choices that mitigate the potential downsides of this linearization. Experiments compare our approach with a wide range of baselines on the molecule generation task and show that our method is more successful at matching the statistics of the original dataset on semantically important metrics. Furthermore, we show that by using appropriate shaping of the latent space, our model allows us to design molecules that are (locally) optimal in desired properties.

arXiv comment8 pages, 5 figures

References in corpus (8)

• Statistical mechanics of complex networks
• Auto-Encoding Variational Bayes
• Semi-Supervised Classification with Graph Convolutional Networks
• Neural Message Passing for Quantum Chemistry
• Junction Tree Variational Autoencoder for Molecular Graph Generation
• Kronecker Graphs: An Approach to Modeling Networks
• Learning Deep Generative Models of Graphs
• Tackling Over-pruning in Variational Autoencoders

Cited by in corpus (61)

• Graph Self-Supervised Learning: A Survey
• A Comprehensive Survey on Deep Graph Representation Learning
• A Gentle Introduction to Deep Learning for Graphs
• SELFIES and the future of molecular string representations
• Perspective on integrating machine learning into computational chemistry and materials science
• Molecular graph generation with Graph Neural Networks
• Deep Learning and Knowledge-Based Methods for Computer Aided Molecular Design -- Toward a Unified Approach: State-of-the-Art and Future Directions
• Hierarchical Generation of Molecular Graphs using Structural Motifs
• Multi-Objective Molecule Generation using Interpretable Substructures
• Machine Learning on Graphs: A Model and Comprehensive Taxonomy
• Augmenting Genetic Algorithms with Deep Neural Networks for Exploring the Chemical Space
• Deep learning of thermodynamics-aware reduced-order models from data
• SchNetPack 2.0: A neural network toolbox for atomistic machine learning
• D-VAE: A Variational Autoencoder for Directed Acyclic Graphs
• Symmetry-adapted generation of 3d point sets for the targeted discovery of molecules
• A Graph to Graphs Framework for Retrosynthesis Prediction
• Learning to Represent Edits
• Iterative Refinement Graph Neural Network for Antibody Sequence-Structure Co-design
• A Deep Generative Model for Fragment-Based Molecule Generation
• Hypergraph Convolution and Hypergraph Attention
• Learning to Extend Molecular Scaffolds with Structural Motifs
• Guiding Deep Molecular Optimization with Genetic Exploration
• Edge-based sequential graph generation with recurrent neural networks
• GraphDF: A Discrete Flow Model for Molecular Graph Generation
• Hierarchical Graph-to-Graph Translation for Molecules
• A Contrastive Variational Graph Auto-Encoder for Node Clustering
• 3DMolNet: A Generative Network for Molecular Structures
• Keep It Simple: Graph Autoencoders Without Graph Convolutional Networks
• A Model to Search for Synthesizable Molecules
• Reinforcement Learning for Molecular Design Guided by Quantum Mechanics
• Local Augmentation for Graph Neural Networks
• Cold Start Similar Artists Ranking with Gravity-Inspired Graph Autoencoders
• A generative model for molecule generation based on chemical reaction trees
• C5T5: Controllable Generation of Organic Molecules with Transformers
• Categorical Normalizing Flows via Continuous Transformations
• Variational Autoencoder for Anti-Cancer Drug Response Prediction
• Graph Deconvolutional Generation
• MIMOSA: Multi-constraint Molecule Sampling for Molecule Optimization
• Disentangling Interpretable Generative Parameters of Random and Real-World Graphs
• Discovering Molecular Functional Groups Using Graph Convolutional Neural Networks
• Adversarial Learned Molecular Graph Inference and Generation
• A Generative Model for Molecular Distance Geometry
• Symmetry-Aware Actor-Critic for 3D Molecular Design
• Spatial Graph Attention and Curiosity-driven Policy for Antiviral Drug Discovery
• Mask-GVAE: Blind Denoising Graphs via Partition
• Learning Latent Space Energy-Based Prior Model for Molecule Generation
• CCGG: A Deep Autoregressive Model for Class-Conditional Graph Generation
• Embeddings and Representation Learning for Structured Data
• Artificial Intelligence based Autonomous Molecular Design for Medical Therapeutic: A Perspective
• CORE: Automatic Molecule Optimization Using Copy & Refine Strategy
• Discrete Object Generation with Reversible Inductive Construction
• Neural Variational Inference For Estimating Uncertainty in Knowledge Graph Embeddings
• Semi-Supervised Hierarchical Drug Embedding in Hyperbolic Space
• Partition and Code: learning how to compress graphs
• Top-N: Equivariant set and graph generation without exchangeability
• Vitruvion: A Generative Model of Parametric CAD Sketches
• MeronymNet: A Hierarchical Approach for Unified and Controllable Multi-Category Object Generation
• Learning to Sample Hard Instances for Graph Algorithms
• Gradient scarcity with Bilevel Optimization for Graph Learning
• Interpreting Equivariant Representations
• Contributions to Representation Learning with Graph Autoencoders and Applications to Music Recommendation