Hierarchical Framework for Retrosynthesis Prediction with Enhanced Reaction Center Localization

TL;DR

This paper introduces a hierarchical AI framework for retrosynthesis that combines reaction center detection, action prediction, and termination, utilizing a contrastively pretrained molecular encoder to improve accuracy and generalization in complex chemical transformations.

Contribution

It presents a novel unified pipeline integrating multiple retrosynthesis tasks with a contrastively pretrained molecular encoder, addressing data scarcity and enhancing prediction robustness.

Findings

Achieves high top-k accuracy on benchmark datasets.

Demonstrates robust reaction center identification.

Effectively handles complex chemical transformations.

Abstract

Retrosynthesis is essential for designing synthetic pathways for complex molecules and can be revolutionized by AI to automate and accelerate chemical synthesis planning for drug discovery and materials science. Here, we propose a hierarchical framework for retrosynthesis prediction that systematically integrates reaction center identification, action prediction, and termination decision into a unified pipeline. Leveraging a molecular encoder pretrained with contrastive learning, the model captures both atom and bond level representations, enabling accurate identification of reaction centers and prediction of chemical actions. The framework addresses the scarcity of multiple reaction center data through augmentation strategies, enhancing the ability of the model to generalize to diverse reaction scenarios. The proposed approach achieves competitive performance across benchmark datasets, with notably high topk accuracy and exceptional reaction center identification capabilities, demonstrating its robustness in handling complex transformations. These advancements position the framework as a promising tool for future applications in material design and drug discovery.