Graph Convolutional Policy Network for Goal-Directed Molecular Graph Generation

Abstract

Generating novel graph structures that optimize given objectives while obeying some given underlying rules is fundamental for chemistry, biology and social science research. This is especially important in the task of molecular graph generation, whose goal is to discover novel molecules with desired properties such as drug-likeness and synthetic accessibility, while obeying physical laws such as chemical valency. However, designing models to find molecules that optimize desired properties while incorporating highly complex and non-differentiable rules remains to be a challenging task. Here we propose Graph Convolutional Policy Network (GCPN), a general graph convolutional network based model for goal-directed graph generation through reinforcement learning. The model is trained to optimize domain-specific rewards and adversarial loss through policy gradient, and acts in an environment that incorporates domain-specific rules. Experimental results show that GCPN can achieve 61% improvement on chemical property optimization over state-of-the-art baselines while resembling known molecules, and achieve 184% improvement on the constrained property optimization task.

arXiv commentNeurIPS 2018, spotlight presentation

References in corpus (13)

• Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
• Inductive Representation Learning on Large Graphs
• Neural Message Passing for Quantum Chemistry
• Molecular Graph Convolutions: Moving Beyond Fingerprints
• Quantum-Chemical Insights from Deep Tensor Neural Networks
• Junction Tree Variational Autoencoder for Molecular Graph Generation
• Objective-Reinforced Generative Adversarial Networks (ORGAN) for Sequence Generation Models
• GraphRNN: Generating Realistic Graphs with Deep Auto-regressive Models
• Learning Deep Generative Models of Graphs
• ChemTS: An Efficient Python Library for de novo Molecular Generation
• Syntax-Directed Variational Autoencoder for Structured Data
• Molecular Generation with Recurrent Neural Networks (RNNs)
• In silico generation of novel, drug-like chemical matter using the LSTM neural network

Cited by in corpus (61)

• Graph Neural Networks: A Review of Methods and Applications
• Recent Advances and Applications of Deep Learning Methods in Materials Science
• Deep learning for molecular design - a review of the state of the art
• GNNExplainer: Generating Explanations for Graph Neural Networks
• Learning Multimodal Graph-to-Graph Translation for Molecular Optimization
• 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
• A Deep Generative Model for Molecule Optimization via One Fragment Modification
• Multi-Objective Molecule Generation using Interpretable Substructures
• Deeply learning molecular structure-property relationships using attention- and gate-augmented graph convolutional network
• Drug-Drug Adverse Effect Prediction with Graph Co-Attention
• Reinforcement Learning for Generative AI: State of the Art, Opportunities and Open Research Challenges
• PGM-Explainer: Probabilistic Graphical Model Explanations for Graph Neural Networks
• All SMILES Variational Autoencoder
• Utilizing Reinforcement Learning for de novo Drug Design
• Learning to Extend Molecular Scaffolds with Structural Motifs
• Learning Graph Neural Networks with Positive and Unlabeled Nodes
• GraphDF: A Discrete Flow Model for Molecular Graph Generation
• Decoding Molecular Graph Embeddings with Reinforcement Learning
• Hierarchical Graph-to-Graph Translation for Molecules
• Finding Patient Zero: Learning Contagion Source with Graph Neural Networks
• GDP: Generalized Device Placement for Dataflow Graphs
• TCL: Transformer-based Dynamic Graph Modelling via Contrastive Learning
• Controllable Data Generation by Deep Learning: A Review
• Differentiable Scaffolding Tree for Molecular Optimization
• Amortized Tree Generation for Bottom-up Synthesis Planning and Synthesizable Molecular Design
• Hit and Lead Discovery with Explorative RL and Fragment-based Molecule Generation
• C5T5: Controllable Generation of Organic Molecules with Transformers
• Uncertainty quantification of molecular property prediction with Bayesian neural networks
• Goal-directed graph construction using reinforcement learning
• Multitask Learning on Graph Neural Networks: Learning Multiple Graph Centrality Measures with a Unified Network
• Graph Deconvolutional Generation
• Towards Interpretable Sparse Graph Representation Learning with Laplacian Pooling
• MIMOSA: Multi-constraint Molecule Sampling for Molecule Optimization
• Black Box Recursive Translations for Molecular Optimization
• Learning Universal Graph Neural Network Embeddings With Aid Of Transfer Learning
• Adversarial Learned Molecular Graph Inference and Generation
• A Generative Model for Molecular Distance Geometry
• Artificial Intelligence in Drug Discovery: Applications and Techniques
• Siamese Graph Neural Networks for Data Integration
• Spatial Graph Attention and Curiosity-driven Policy for Antiviral Drug Discovery
• Few-shot Network Anomaly Detection via Cross-network Meta-learning
• Deep Q-Learning for Directed Acyclic Graph Generation
• SCGG: A Deep Structure-Conditioned Graph Generative Model
• Learning Latent Space Energy-Based Prior Model for Molecule Generation
• Neural Packet Classification
• An In-depth Summary of Recent Artificial Intelligence Applications in Drug Design
• Uncertainty quantification of molecular property prediction using Bayesian neural network models
• A* Search Without Expansions: Learning Heuristic Functions with Deep Q-Networks
• Learning Space Partitions for Path Planning
• A Graph Deep Learning Framework for High-Level Synthesis Design Space Exploration
• LEReg: Empower Graph Neural Networks with Local Energy Regularization
• Permutation-Invariant Variational Autoencoder for Graph-Level Representation Learning
• Deep Learning on Attributed Graphs: A Journey from Graphs to Their Embeddings and Back
• Geometric learning of the conformational dynamics of molecules using dynamic graph neural networks
• D2C: Diffusion-Denoising Models for Few-shot Conditional Generation
• Molecular Attributes Transfer from Non-Parallel Data
• Network representation learning: A macro and micro view
• Learning Mechanically Driven Emergent Behavior with Message Passing Neural Networks
• Transferable Graph Optimizers for ML Compilers
• Transport based Graph Kernels