RetroMPA: A Molecular Property-Aware Auxiliary Framework for Enhancing Retrosynthesis Prediction

2026-08-17

AI Summary

This study addresses the prediction inaccuracies in data-driven retrosynthesis models caused by insufficient chemical priors by proposing RetroMPA, a plug-and-play post-processing framework. RetroMPA introduces a novel model-agnostic chemical filtering augmentation mechanism that calibrates predicted pathways via molecular property-aware latent space alignment. This approach seamlessly integrates chemical knowledge into both template-based and template-free models without requiring retraining. Experiments demonstrate that RetroMPA improves Top-1 accuracy by 5.50% on USPTO-50K and 2.03% on USPTO-Full. Furthermore, wet-lab validation confirms the reaction feasibility of novel substrate combinations identified by the framework. These results indicate that RetroMPA significantly enhances the generalization capability and practical utility of existing retrosynthesis models by effectively bridging the gap between data-driven predictions and fundamental chemical constraints.

Abstract

Retrosynthesis is a cornerstone of drug discovery and organic synthesis. While data-driven deep learning models have shown remarkable progress, they autonomously learn reaction patterns from extensive datasets with limited integration of established chemical knowledge as priors. To address this limitation, we introduce RetroMPA, a molecular property-aware, post-hoc enhancement module that injects chemical knowledge into the retrosynthesis pipeline. Rather than functioning as an independent SMILES sequence generator, RetroMPA is a broadly applicable, model-agnostic chemical filter designed to recalibrate and optimize the predictive pathways of existing algorithms. This plug-and-play framework integrates seamlessly with a range of data-driven retrosynthesis methods, enhancing outputs without modifying model architecture or requiring resource-intensive retraining. By leveraging a property-aware latent embedding space, RetroMPA consistently improves top-1 accuracy across eight representative retrosynthesis models by an average of 5.50% on USPTO-50K. Furthermore, we validate its scalability on the large-scale USPTO-Full dataset, achieving an average improvement of about 2.03% across both template-based and template-free architectures. Wet-lab experiments provide preliminary support for the practical utility of the framework. These syntheses confirmed viable, previously unreported substrate combinations for classic reaction paradigms---specifically, Suzuki-Miyaura coupling, Bucherer reaction, and Friedel-Crafts acylation---suggesting that RetroMPA can operate beyond mere data fitting. The code is open-sourced at https://github.com/MengzhouLu/RetroMPA.

Problem

• Retrosynthesis prediction
• Chemical knowledge integration
• Deep learning
• Molecular property awareness

Innovation

• Retrosynthesis Prediction
• Molecular Property-Aware
• Model-Agnostic
• Plug-and-Play Framework
• Chemical Knowledge Injection