# Learning to Make Generalizable and Diverse Predictions for Retrosynthesis

arXiv: https://arxiv.org/abs/1910.09688 · HTML: https://mlsanity.com/paper/1910.09688/

## Key facts

- Cited by 22 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 21 Oct 2019.

## Built on by (most cited first, 22 of 22)

- [Permutation invariant graph-to-sequence model for template-free retrosynthesis and reaction prediction](https://mlsanity.com/paper/2110.09681/) (2021)
- [FusionRetro: Molecule Representation Fusion via In-Context Learning for Retrosynthetic Planning](https://mlsanity.com/paper/2209.15315/) (2022)
- [Modern Hopfield Networks for Few- and Zero-Shot Reaction Template Prediction](https://mlsanity.com/paper/2104.03279/) (2021)
- [SemiRetro: Semi-template framework boosts deep retrosynthesis prediction](https://mlsanity.com/paper/2202.08205/) (2022)
- [DirectMultiStep: Direct Route Generation for Multistep Retrosynthesis](https://mlsanity.com/paper/2405.13983/) (2024)
- [Deep Learning Methods for Small Molecule Drug Discovery: A Survey](https://mlsanity.com/paper/2303.00313/) (2023)
- [Node-Aligned Graph-to-Graph (NAG2G): Elevating Template-Free Deep Learning Approaches in Single-Step Retrosynthesis](https://mlsanity.com/paper/2309.15798/) (2023)
- [Retroformer: Pushing the Limits of Interpretable End-to-end Retrosynthesis Transformer](https://mlsanity.com/paper/2201.12475/) (2022)
- [MARS: A Motif-based Autoregressive Model for Retrosynthesis Prediction](https://mlsanity.com/paper/2209.13178/) (2022)
- [RetroDiff: Retrosynthesis as Multi-stage Distribution Interpolation](https://mlsanity.com/paper/2311.14077/) (2023)
- [RetCL: A Selection-based Approach for Retrosynthesis via Contrastive Learning](https://mlsanity.com/paper/2105.00795/) (2021)
- [Preference Optimization for Molecule Synthesis with Conditional Residual Energy-based Models](https://mlsanity.com/paper/2406.02066/) (2024)
- [RETRO SYNFLOW: Discrete Flow Matching for Accurate and Diverse Single-Step Retrosynthesis](https://mlsanity.com/paper/2506.04439/) (2025)
- [Recent advances in artificial intelligence for retrosynthesis](https://mlsanity.com/paper/2301.05864/) (2023)
- [mathsfG2Retro as a Two-Step Graph Generative Models for Retrosynthesis Prediction](https://mlsanity.com/paper/2206.04882/) (2022)
- [A Unified View of Deep Learning for Reaction and Retrosynthesis Prediction: Current Status and Future Challenges](https://mlsanity.com/paper/2306.15890/) (2023)
- [Modeling Diverse Chemical Reactions for Single-step Retrosynthesis via Discrete Latent Variables](https://mlsanity.com/paper/2208.05482/) (2022)
- [Survey on Recent Progress of AI for Chemistry: Methods, Applications, and Opportunities](https://mlsanity.com/paper/2502.17456/) (2025)
- [MotifRetro: Exploring the Combinability-Consistency Trade-offs in retrosynthesis via Dynamic Motif Editing](https://mlsanity.com/paper/2305.15153/) (2023)
- [Evolutionary Retrosynthetic Route Planning](https://mlsanity.com/paper/2310.05186/) (2023)
- [G-MATT: Single-step Retrosynthesis Prediction using Molecular Grammar Tree Transformer](https://mlsanity.com/paper/2305.03153/) (2023)
- [Coverage Before Control: Route-Instruction Grounding and Steering for Controllable Retrosynthesis](https://mlsanity.com/paper/2609.39955/) (2026)

## Abstract (arXiv)

We propose a new model for making generalizable and diverse retrosynthetic reaction predictions. Given a target compound, the task is to predict the likely chemical reactants to produce the target. This generative task can be framed as a sequence-to-sequence problem by using the SMILES representations of the molecules. Building on top of the popular Transformer architecture, we propose two novel pre-training methods that construct relevant auxiliary tasks (plausible reactions) for our problem. Furthermore, we incorporate a discrete latent variable model into the architecture to encourage the model to produce a diverse set of alternative predictions. On the 50k subset of reaction examples from the United States patent literature (USPTO-50k) benchmark dataset, our model greatly improves performance over the baseline, while also generating predictions that are more diverse.