Reaction Prediction

This group covers models that predict the products of chemical reactions (forward prediction) or propose synthetic routes to target molecules (retrosynthesis) by treating reactions as sequence-to-sequence translation problems over SMILES strings.

Paper	Year	Task	Key Idea
NMT Reaction Prediction	2016	Forward	GRU seq2seq model translating reactants to products
Molecular Transformer	2019	Forward	Attention-based seq2seq achieving 90.4% top-1 accuracy with calibration
Data Transfer Retrosynthesis	2020	Retrosynthesis	Pre-training on USPTO-Full improves retrosynthesis accuracy
Tied Two-Way Transformer	2021	Retrosynthesis	Shared parameters with cycle consistency for diverse predictions
ReactionT5	2023	Forward + Retro	Two-stage T5 pretraining for minimal fine-tuning data

All Notes

Computational Chemistry

Three data transfer methods for retrosynthesis: pre-training plus fine-tuning, multi-task learning, and self-training

Data Transfer Approaches for Seq-to-Seq Retrosynthesis

A systematic study of data transfer techniques (joint training, self-training, pre-training plus fine-tuning) applied to Transformer-based retrosynthesis. Pre-training on USPTO-Full followed by fine-tuning on USPTO-50K achieves the best results, improving top-1 accuracy from 35.3% to 57.4%.

Computational Chemistry

Seq2seq encoder-decoder translating reactant SMILES to product SMILES for reaction prediction

Neural Machine Translation for Reaction Prediction

This 2016 paper first proposed treating organic reaction prediction as a neural machine translation problem, using a GRU-based sequence-to-sequence model with attention to translate reactant SMILES strings into product SMILES strings.

Computational Chemistry

ReactionT5 two-stage pretraining from CompoundT5 to ReactionT5 with product prediction and yield results

ReactionT5: Pre-trained T5 for Reaction Prediction

ReactionT5 introduces a two-stage pretraining pipeline (compound then reaction) on the Open Reaction Database, enabling competitive product and yield prediction with as few as 30 fine-tuning reactions.

Computational Chemistry

Diagram of the tied two-way transformer architecture with shared encoder, retro and forward decoders, latent variables, and cycle consistency, alongside USPTO-50K accuracy and validity results

Tied Two-Way Transformers for Diverse Retrosynthesis

This paper couples a retrosynthesis transformer with a forward reaction transformer through parameter sharing, cycle consistency checks, and multinomial latent variables. The combined approach reduces top-1 SMILES invalidity to 0.1% on USPTO-50K, improves top-10 accuracy to 78.5%, and achieves 87.3% pathway coverage on a multi-pathway in-house dataset.

Computational Chemistry

Molecular Transformer architecture showing atom-wise tokenized SMILES input through encoder-decoder with multi-head attention to predict reaction products

Molecular Transformer: Calibrated Reaction Prediction

The Molecular Transformer applies the Transformer architecture to forward reaction prediction, treating it as SMILES-to-SMILES machine translation. It achieves 90.4% top-1 accuracy on USPTO_MIT, outperforms quantum-chemistry baselines on regioselectivity, and provides calibrated uncertainty scores (0.89 AUC-ROC) for ranking synthesis pathways.

All Notes#

All Notes