Deep learning for molecular design - a review of the state of the art

Abstract

In the space of only a few years, deep generative modeling has revolutionized how we think of artificial creativity, yielding autonomous systems which produce original images, music, and text. Inspired by these successes, researchers are now applying deep generative modeling techniques to the generation and optimization of molecules - in our review we found 45 papers on the subject published in the past two years. These works point to a future where such systems will be used to generate lead molecules, greatly reducing resources spent downstream synthesizing and characterizing bad leads in the lab. In this review we survey the increasingly complex landscape of models and representation schemes that have been proposed. The four classes of techniques we describe are recursive neural networks, autoencoders, generative adversarial networks, and reinforcement learning. After first discussing some of the mathematical fundamentals of each technique, we draw high level connections and comparisons with other techniques and expose the pros and cons of each. Several important high level themes emerge as a result of this work, including the shift away from the SMILES string representation of molecules towards more sophisticated representations such as graph grammars and 3D representations, the importance of reward function design, the need for better standards for benchmarking and testing, and the benefits of adversarial training and reinforcement learning over maximum likelihood based training.

arXiv comment24 pages, new title, published in RSC MSDE

References in corpus (45)

• Improving neural networks by preventing co-adaptation of feature detectors
• GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium
• Categorical Reparameterization with Gumbel-Softmax
• Automatic chemical design using a data-driven continuous representation of molecules
• Learning to Plan Chemical Syntheses
• Improved Techniques for Training GANs
• Deep Reinforcement Learning for De-Novo Drug Design
• GuacaMol: Benchmarking Models for De Novo Molecular Design
• Junction Tree Variational Autoencoder for Molecular Graph Generation
• Machine learning prediction errors better than DFT accuracy
• Tensor field networks: Rotation- and translation-equivariant neural networks for 3D point clouds
• Objective-Reinforced Generative Adversarial Networks (ORGAN) for Sequence Generation Models
• MolGAN: An implicit generative model for small molecular graphs
• Learning Deep Generative Models of Graphs
• Deep Reinforcement Learning that Matters
• SMILES Enumeration as Data Augmentation for Neural Network Modeling of Molecules
• ChemTS: An Efficient Python Library for de novo Molecular Generation
• Applying machine learning techniques to predict the properties of energetic materials
• Microstructural Materials Design via Deep Adversarial Learning Methodology
• Graph Convolutional Policy Network for Goal-Directed Molecular Graph Generation
• Syntax-Directed Variational Autoencoder for Structured Data
• Conditional molecular design with deep generative models
• Improving Chemical Autoencoder Latent Space and Molecular De novo Generation Diversity with Heteroencoders
• Maximum-Likelihood Augmented Discrete Generative Adversarial Networks
• Multi-task Neural Networks for QSAR Predictions
• Grammar Variational Autoencoder
• Chemception: A Deep Neural Network with Minimal Chemistry Knowledge Matches the Performance of Expert-developed QSAR/QSPR Models
• A Universal Density Matrix Functional from Molecular Orbital-Based Machine Learning: Transferability across Organic Molecules
• ChemGAN challenge for drug discovery: can AI reproduce natural chemical diversity?
• Population-based de novo molecule generation, using grammatical evolution
• Wavelet Scattering Regression of Quantum Chemical Energies
• Constrained Generation of Semantically Valid Graphs via Regularizing Variational Autoencoders
• Tuning the Molecular Weight Distribution from Atom Transfer Radical Polymerization Using Deep Reinforcement Learning
• Constrained Bayesian Optimization for Automatic Chemical Design
• Molecular Hypergraph Grammar with its Application to Molecular Optimization
• On the State of the Art of Evaluation in Neural Language Models
• Molecular Generation with Recurrent Neural Networks (RNNs)
• In silico generation of novel, drug-like chemical matter using the LSTM neural network
• Towards GAN Benchmarks Which Require Generalization
• Adversarial Autoencoders with Constant-Curvature Latent Manifolds
• Decoding Molecular Graph Embeddings with Reinforcement Learning
• The Chemical Space of B, N-substituted Polycyclic Aromatic Hydrocarbons: Combinatorial Enumeration and High-Throughput First-Principles Modeling
• Machine Learning of Energetic Material Properties
• Actively Learning what makes a Discrete Sequence Valid
• Independent Vector Analysis for Data Fusion Prior to Molecular Property Prediction with Machine Learning

Cited by in corpus (74)

• Self-Referencing Embedded Strings (SELFIES): A 100% robust molecular string representation
• Big-Data Science in Porous Materials: Materials Genomics and Machine Learning
• Machine learning for electronically excited states of molecules
• Autonomous discovery in the chemical sciences part I: Progress
• Graph Representation Learning in Biomedicine
• Generative adversarial networks (GAN) based efficient sampling of chemical space for inverse design of inorganic materials
• Autonomous discovery in the chemical sciences part II: Outlook
• Perspective on integrating machine learning into computational chemistry and materials science
• Inverse design of 3d molecular structures with conditional generative neural networks
• Exploring Chemical Space using Natural Language Processing Methodologies for Drug Discovery
• Generative chemistry: drug discovery with deep learning generative models
• Regression Transformer: Concurrent sequence regression and generation for molecular language modeling
• Deep Learning and Knowledge-Based Methods for Computer Aided Molecular Design -- Toward a Unified Approach: State-of-the-Art and Future Directions
• Diffusion Models in $\text{De Novo}$ Drug Design
• Machine Learning on Graphs: A Model and Comprehensive Taxonomy
• Graph Neural Networks for Molecules
• Graph-augmented Convolutional Networks on Drug-Drug Interactions Prediction
• PubChemQC PM6: A dataset of 221 million molecules with optimized molecular geometries and electronic properties
• High-throughput property-driven generative design of functional organic molecules
• Machine learning-assisted design of material properties
• Physics-Informed Neural Networks for Nonhomogeneous Material Identification in Elasticity Imaging
• Scaffold-constrained molecular generation
• Physically inspired deep learning of molecular excitations and photoemission spectra
• Machine learning and excited-state molecular dynamics
• We Should at Least Be Able to Design Molecules That Dock Well
• Learning To Navigate The Synthetically Accessible Chemical Space Using Reinforcement Learning
• Constrained Bayesian Optimization for Automatic Chemical Design
• Using Molecular Embeddings in QSAR Modeling: Does it Make a Difference?
• Generative Language Model for Catalyst Discovery
• CRYSPNet: Crystal Structure Predictions via Neural Network
• Sample-Efficient Optimization in the Latent Space of Deep Generative Models via Weighted Retraining
• Graph Machine Learning for Design of High-Octane Fuels
• The Expressive Power of Graph Neural Networks: A Survey
• Graph neural networks for the prediction of molecular structure-property relationships
• From Stars to Subgraphs: Uplifting Any GNN with Local Structure Awareness
• Controllable Data Generation by Deep Learning: A Review
• Acoustic Cloak Design via Machine Learning
• Building powerful and equivariant graph neural networks with structural message-passing
• Faster and more diverse de novo molecular optimization with double-loop reinforcement learning using augmented SMILES
• Amortized Tree Generation for Bottom-up Synthesis Planning and Synthesizable Molecular Design
• Learning 3D Representations of Molecular Chirality with Invariance to Bond Rotations
• Message Passing Networks for Molecules with Tetrahedral Chirality
• Graph Neural Networks for the Prediction of Substrate-Specific Organic Reaction Conditions
• Molecular-orbital-based Machine Learning for Open-shell and Multi-reference Systems with Kernel Addition Gaussian Process Regression
• Automated patent extraction powers generative modeling in focused chemical spaces
• Towards Better Opioid Antagonists Using Deep Reinforcement Learning
• Discovering Molecular Functional Groups Using Graph Convolutional Neural Networks
• IMG2SMI: Translating Molecular Structure Images to Simplified Molecular-input Line-entry System
• Artificial Intelligence in Drug Discovery: Applications and Techniques
• Molecular Dipole Moment Learning via Rotationally Equivariant Gaussian Process Regression with Derivatives in Molecular-orbital-based Machine Learning
• Counting Substructures with Higher-Order Graph Neural Networks: Possibility and Impossibility Results
• Predicting antimicrobial activity of conjugated oligoelectrolyte molecules via machine learning
• pyscreener: A Python Wrapper for Computational Docking Software
• Virtual Reality for Understanding Artificial-Intelligence-driven Scientific Discovery with an Application in Quantum Optics
• G-PCGRL: Procedural Graph Data Generation via Reinforcement Learning
• Evaluating the progress of Deep Reinforcement Learning in the real world: aligning domain-agnostic and domain-specific research
• Prospecting for Pluripotency in Metamaterial Design
• Surrogate-Based Black-Box Optimization Method for Costly Molecular Properties
• Characterizing the Latent Space of Molecular Deep Generative Models with Persistent Homology Metrics
• Physics-Constrained Predictive Molecular Latent Space Discovery with Graph Scattering Variational Autoencoder
• An In-depth Summary of Recent Artificial Intelligence Applications in Drug Design
• Artificial Intelligence based Autonomous Molecular Design for Medical Therapeutic: A Perspective
• Constrained Discrete Black-Box Optimization using Mixed-Integer Programming
• Applying Deutsch's concept of good explanations to artificial intelligence and neuroscience -- an initial exploration
• Scaffold Embeddings: Learning the Structure Spanned by Chemical Fragments, Scaffolds and Compounds
• Graph Random Neural Features for Distance-Preserving Graph Representations
• Augmenting Molecular Images with Vector Representations as a Featurization Technique for Drug Classification
• Multi-Constraint Molecular Generation using Sparsely Labelled Training Data for Localized High-Concentration Electrolyte Diluent Screening
• Graph-Convolutional Deep Learning to Identify Optimized Molecular Configurations
• Inverse Structural Design of Graphene/Boron Nitride Hybrids by Regressional GAN
• Continuous Representation of Molecules Using Graph Variational Autoencoder
• Scaffold-Induced Molecular Graph (SIMG): Effective Graph Sampling Methods for High-Throughput Computational Drug Discovery
• Autoencoding Undirected Molecular Graphs With Neural Networks
• Quantum dynamics evolution predicted by the long short-term memory network in the photosystem II reaction center