Retrosynthetic Reaction Prediction Using Neural Sequence-to-Sequence Models

By Bowen Liu, Bharath Ramsundar, Prasad Kawthekar, Jade Shi, Joseph Gomes, Quang Luu Nguyen, Stephen Ho, J. Sloane, P. Wender, Vijay S. Pande. 2017

486 citations recorded by the source corpus.

We describe a fully data driven model that learns to perform a retrosynthetic reaction prediction task, which is treated as a sequence-to-sequence mapping problem. The end-to-end trained model has an encoder–decoder architecture that consists of two recurrent neural networks, which has previously shown great success in solving other sequence-to-sequence prediction tasks such as machine translation. The model is trained on 50,000 experimental reaction examples from the United States patent literature, which span 10 broad reaction types that are commonly used by medicinal chemists. We find that our model performs comparably with a rule-based expert system baseline model, and also overcomes certain limitations associated with rule-based expert systems and with any machine learning approach that contains a rule-based expert system component. Our model provides an important first step toward solving the challenging problem of computational retrosynthetic analysis.

Subfields: Control and Systems Engineering, Materials Chemistry, Spectroscopy, Model Architectures, Computational Chemistry & Drug Discovery

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Control and Systems Engineering

1. Gisbert Schneider
2. Yue Zhao
3. Aboul Ella Otifey Hassanien
4. Jure Leskovec
5. Jimeng Sun
6. José Jiménez-Luna
7. Sepp Hochreiter
8. Jörg Kurt Wegner

In Materials Chemistry

1. Gisbert Schneider
2. Francesca Grisoni
3. Jure Leskovec
4. Sepp Hochreiter
5. Marinka Zitnik
6. Philippe Schwaller
7. Günter Klambauer
8. Rex Ying

In Spectroscopy

1. Brooke E. Husic
2. Jörg Kurt Wegner
3. C. Bekas
4. Connor W. Coley
5. Philippe Schwaller
6. Rohit Batra
7. Gisbert Schneider
8. Cao Xiao

In Model Architectures

1. Jure Leskovec
2. Rex Ying
3. Günter Klambauer
4. Jimeng Sun
5. Hongming Chen
6. Ola Engkvist
7. Gisbert Schneider
8. Sepp Hochreiter

In Computational Chemistry & Drug Discovery

1. Jure Leskovec
2. Sepp Hochreiter
3. Günter Klambauer
4. Rex Ying
5. Jiaxuan You
6. Gisbert Schneider
7. Francesca Grisoni
8. Philippe Schwaller

Citations in the corpus

• Graph Convolutional Policy Network for Goal-Directed Molecular Graph Generation (2018, background): “For example, a single character perturbation in a text-based representation of a molecule can lead to significant changes to the underlying molecular structure or even invalidate it [30].”
• Unsupervised learning of dynamical and molecular similarity using variance minimization (2017, methodology): “To illustrate this, we ran a separate analysis of chemical reactions obtained from the United States patent literature and processed as described in [Liu et al., 2017] for the prediction of reaction products.”
• “Found in Translation”: predicting outcomes of complex organic chemistry reactions using neural sequence-to-sequence models (2017, methodology): “This is a considerably more difficult task for a seq2seq model and was approached by Liu et al. [24].”
• Emerging materials intelligence ecosystems propelled by machine learning (2020, result): “There have also been recent attempts to directly predict the SMILES representation of the reactants 215,216 (or the products, for the forward reaction prediction problem 217,218 ).”
• The rise of deep learning in drug discovery. (2018, background): “Liu et al. [78] used neural sequence-to-sequence models for retrosynthetic prediction.”