Geometric deep learning on molecular representations (2021-07-26T00:00:00.000000Z)

TL;DR

This Review provides a structured and harmonized overview of molecular GDL, highlighting its applications in drug discovery, chemical synthesis prediction and quantum chemistry, as well as relevant molecular representations, such as molecular graphs, grids, surfaces and strings.

Abstract

Geometric deep learning (GDL) is based on neural network architectures that incorporate and process symmetry information. GDL bears promise for molecular modelling applications that rely on molecular representations with different symmetry properties and levels of abstraction. This Review provides a structured and harmonized overview of molecular GDL, highlighting its applications in drug discovery, chemical synthesis prediction and quantum chemistry. It contains an introduction to the principles of GDL, as well as relevant molecular representations, such as molecular graphs, grids, surfaces and strings, and their respective properties. The current challenges for GDL in the molecular sciences are discussed, and a forecast of future opportunities is attempted. Geometric representations are becoming more important in molecular deep learning as the spatial structure of molecules contains important information about their properties. Kenneth Atz and colleagues review current progress and challenges in this emerging field of geometric deep learning.

Authors

F. Grisoni

3 papers

G. Schneider

3 papers

Kenneth Atz

1 papers

TL;DR

Abstract

Authors

References209 items

Inverse design of 3d molecular structures with conditional generative neural networks

Geometric deep learning of RNA structure

Chemical language models enable navigation in sparsely populated chemical space

Highly accurate protein structure prediction with AlphaFold

QMugs, quantum mechanical properties of drug-like molecules

Accurate prediction of protein structures and interactions using a 3-track neural network

MoCL: Data-driven Molecular Fingerprint via Knowledge-aware Contrastive Learning from Molecular Graph

SE(3)-equivariant prediction of molecular wavefunctions and electronic densities

SpookyNet: Learning force fields with electronic degrees of freedom and nonlocal effects

Geometric Deep Learning: Grids, Groups, Graphs, Geodesics, and Gauges

MPGVAE: improved generation of small organic molecules using message passing neural nets

Artificial intelligence in drug discovery: recent advances and future perspectives

Accelerated antimicrobial discovery via deep generative models and molecular dynamics simulations

Weisfeiler and Lehman Go Topological: Message Passing Simplicial Networks

E(n) Equivariant Graph Neural Networks

Spherical Message Passing for 3D Molecular Graphs

OctSurf: Efficient hierarchical voxel-based molecular surface representation for protein-ligand affinity prediction.

Equivariant message passing for the prediction of tensorial properties and molecular spectra

E(3)-equivariant graph neural networks for data-efficient and accurate interatomic potentials

Combining generative artificial intelligence and on-chip synthesis for de novo drug design

LieTransformer: Equivariant self-attention for Lie Groups

Molecular machine learning with conformer ensembles

Fast and Uncertainty-Aware Directional Message Passing for Non-Equilibrium Molecules

Data augmentation strategies to improve reaction yield predictions and estimate uncertainty

Coloring Molecules with Explainable Artificial Intelligence for Preclinical Relevance Assessment

Molecular Mechanics-Driven Graph Neural Network with Multiplex Graph for Molecular Structures

I Overview

Predicting enzymatic reactions with a molecular transformer

DECIMER: towards deep learning for chemical image recognition

Predicting Chemical Properties using Self-Attention Multi-task Learning based on SMILES Representation

ChemBERTa: Large-Scale Self-Supervised Pretraining for Molecular Property Prediction

Quantum chemical accuracy from density functional approximations via machine learning

Machine Learning Force Fields

Fermentative production of Vitamin E tocotrienols in Saccharomyces cerevisiae under cold-shock-triggered temperature control

Learning Mesh-Based Simulation with Graph Networks

Quantum based machine learning of competing chemical reaction profiles

Transfer learning enables the molecular transformer to predict regio- and stereoselective reactions on carbohydrates

NURA: A curated dataset of nuclear receptor modulators.

Molecular optimization by capturing chemist’s intuition using deep neural networks

Quantum machine learning using atom-in-molecule-based fragments selected on the fly

Euclidean Symmetry and Equivariance in Machine Learning

Relevance of Rotationally Equivariant Convolutions for Predicting Molecular Properties

Optimizing Pharmacokinetic Property Prediction Based on Integrated Datasets and a Deep Learning Approach

Mapping the space of chemical reactions using attention-based neural networks

Prediction of chemical reaction yields using deep learning

Finding Symmetry Breaking Order Parameters with Euclidean Neural Networks

Drug discovery with explainable artificial intelligence

SE(3)-Transformers: 3D Roto-Translation Equivariant Attention Networks

Self-Supervised Graph Transformer on Large-Scale Molecular Data

Improving Graph Neural Network Expressivity via Subgraph Isomorphism Counting

Learning Graph Models for Retrosynthesis Prediction

Learning Graph Models for Template-Free Retrosynthesis

Amino acid encoding for deep learning applications

XAI for Graphs: Explaining Graph Neural Network Predictions by Identifying Relevant Walks

Thousands of reactants and transition states for competing E2 and S N 2 reactions

Language Models are Few-Shot Learners

Open Graph Benchmark: Datasets for Machine Learning on Graphs

Learning molecular dynamics with simple language model built upon long short-term memory neural network

Let's Agree to Degree: Comparing Graph Convolutional Networks in the Message-Passing Framework

A Deep Learning Approach to Antibiotic Discovery

Directional Message Passing for Molecular Graphs

State-of-the-art augmented NLP transformer models for direct and single-step retrosynthesis

Predicting retrosynthetic pathways using transformer-based models and a hyper-graph exploration strategy

A self-attention based message passing neural network for predicting molecular lipophilicity and aqueous solubility

DeepSurf: A surface-based deep learning approach for the prediction of ligand binding sites on proteins

Can graph neural networks count substructures?

Multi-Objective Molecule Generation using Interpretable Substructures

Exploring Chemical Space using Natural Language Processing Methodologies for Drug Discovery

Predicting Binding from Screening Assays with Transformer Network Embeddings

Building attention and edge message passing neural networks for bioactivity and physical–chemical property prediction

Bidirectional Molecule Generation with Recurrent Neural Networks

Deciphering interaction fingerprints from protein molecular surfaces using geometric deep learning

Transformer neural network for protein-specific de novo drug generation as a machine translation problem

Exploring chemical compound space with quantum-based machine learning

Constrained Bayesian optimization for automatic chemical design using variational autoencoders

A novel molecular representation with BiGRU neural networks for learning atom