Permutation invariant graph-to-sequence model for template-free retrosynthesis and reaction prediction (2021-10-19T00:00:00.000000Z)

TL;DR

A novel Graph2SMILES model that combines the power of Transformer models for text generation with the permutation invariance of molecular graph encoders that mitigates the need for input data augmentation is described.

Abstract

Synthesis planning and reaction outcome prediction are two fundamental problems in computer-aided organic chemistry for which a variety of data-driven approaches have emerged. Natural language approaches that model each problem as a SMILES-to-SMILES translation lead to a simple end-to-end formulation, reduce the need for data preprocessing, and enable the use of well-optimized machine translation model architectures. However, SMILES representations are not efficient for capturing information about molecular structures, as evidenced by the success of SMILES augmentation to boost empirical performance. Here, we describe a novel Graph2SMILES model that combines the power of Transformer models for text generation with the permutation invariance of molecular graph encoders that mitigates the need for input data augmentation. In our encoder, a directed message passing neural network (D-MPNN) captures local chemical environments, and the global attention encoder allows for long-range and intermolecular interactions, enhanced by graph-aware positional embedding. As an end-to-end architecture, Graph2SMILES can be used as a drop-in replacement for the Transformer in any task involving molecule(s)-to-molecule(s) transformations, which we empirically demonstrate leads to improved performance on existing benchmarks for both retrosynthesis and reaction outcome prediction.

Authors

Connor W. Coley

8 papers

Zhengkai Tu

1 papers

TL;DR

Abstract

Authors

References77 items

Deep Retrosynthetic Reaction Prediction using Local Reactivity and Global Attention

Chemformer: a pre-trained transformer for computational chemistry

Dual-view Molecule Pre-training

Do Transformers Really Perform Bad for Graph Representation?

Non-Autoregressive Electron Redistribution Modeling for Reaction Prediction

How Attentive are Graph Attention Networks?

GTA: Graph Truncated Attention for Retrosynthesis

RetroPrime: A Diverse, Plausible and Transformer-based Method for Single-Step Retrosynthesis Predictions

Meta Learning for Low-Resource Molecular Optimization

Valid, Plausible, and Diverse Retrosynthesis Using Tied Two-Way Transformers with Latent Variables

PCT: Point cloud transformer

Energy-based View of Retrosynthesis

Molecule Edit Graph Attention Network: Modeling Chemical Reactions as Sequences of Graph Edits

Self-Supervised Graph Transformer on Large-Scale Molecular Data

RetroXpert: Decompose Retrosynthesis Prediction like a Chemist

Learning Graph Models for Template-Free Retrosynthesis

Graph-Aware Transformer: Is Attention All Graphs Need?

A Graph to Graphs Framework for Retrosynthesis Prediction

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

Normalization

Molecule Attention Transformer

Integrating Deep Neural Networks and Symbolic Inference for Organic Reactivity Prediction

Retrosynthesis with attention-based NMT model and chemical analysis of “wrong” predictions

Retrosynthesis Prediction with Conditional Graph Logic Network

Predicting Retrosynthetic Reactions Using Self-Corrected Transformer Neural Networks

Graph Transformer for Graph-to-Sequence Learning

Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer

Molecular Transformer unifies reaction prediction and retrosynthesis across pharma chemical space.

Molecular Graph Enhanced Transformer for Retrosynthesis Prediction

Learning to Make Generalizable and Diverse Predictions for Retrosynthesis

Self-Attention with Structural Position Representations

Retrosynthesis with Attention-Based NMT Model and Chemical Analysis of the "Wrong" Predictions

Strategies for Pre-training Graph Neural Networks

Path-Augmented Graph Transformer Network

Automatic Retrosynthetic Pathway Planning Using Template-free Models

RDChiral: An RDKit Wrapper for Handling Stereochemistry in Retrosynthetic Template Extraction and Application

Analyzing Learned Molecular Representations for Property Prediction

Enhancing Retrosynthetic Reaction Prediction with Deep Learning Using Multiscale Reaction Classification

Transformer-XL: Attentive Language Models beyond a Fixed-Length Context

Graph Transformation Policy Network for Chemical Reaction Prediction

Molecular Transformer: A Model for Uncertainty-Calibrated Chemical Reaction Prediction

A graph-convolutional neural network model for the prediction of chemical reactivity

SQL-to-Text Generation with Graph-to-Sequence Model

A Generative Model For Electron Paths

Graph2Seq: Graph to Sequence Learning with Attention-based Neural Networks

Self-Attention with Relative Position Representations

Junction Tree Variational Autoencoder for Molecular Graph Generation

Computer-Assisted Retrosynthesis Based on Molecular Similarity

Graph Attention Networks

Predicting Organic Reaction Outcomes with Weisfeiler-Lehman Network

Attention is All you Need

Retrosynthetic Reaction Prediction Using Neural Sequence-to-Sequence Models

Neural-Symbolic Machine Learning for Retrosynthesis and Reaction Prediction.

SMILES Enumeration as Data Augmentation for Neural Network Modeling of Molecules

OpenNMT: Open-Source Toolkit for Neural Machine Translation

Linking the Neural Machine Translation and the Prediction of Organic Chemistry Reactions

Bridging Nonlinearities and Stochastic Regularizers with Gaussian Error Linear Units

Computer-Assisted Synthetic Planning: The End of the Beginning.

Adam: A Method for Stochastic Optimization

Neural Machine Translation by Jointly Learning to Align and Translate

Learning Phrase Representations using RNN Encoder–Decoder for Statistical Machine Translation

Extraction of chemical structures and reactions from the literature

Extended-Connectivity Fingerprints

Assigning Unique Keys to Chemical Compounds for Data Integration: Some Interesting Counter Examples

Extensions of marginalized graph kernels

On the All-Pairs-Shortest-Path Problem in Unweighted Undirected Graphs

SMILES, a chemical language and information system. 1. Introduction to methodology and encoding rules

Computer-assisted design of complex organic syntheses.

Geometry-aware Transformer for molecular property prediction

As an illustration of how Graph2SMILES can benefit from additional features or techniques, we briefly experiment with one of them, latent variable modeling similar to Chen et al

BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding

: Open-source cheminformatics

The logic of chemical synthesis

Robert Robinson Lecture. Retrosynthetic thinking—essentials and examples

Computer-assisted synthetic analysis for complex molecules. Methods and procedures for machine generation of synthetic intermediates

Graph2SMILES (D-GCN) (ours)