ReactionTeam: Teaming Experts for Divergent Thinking Beyond Typical Reaction Patterns

TL;DR

ReactionTeam introduces a novel ensemble of expert models that captures diverse reaction outcomes in chemical synthesis, addressing the limitations of likelihood-based models by embracing the stochastic nature of reactions to improve prediction accuracy.

Contribution

The paper presents ReactionTeam, a framework of specialized expert models that generate and rank diverse plausible reaction outcomes, advancing reaction prediction beyond common patterns.

Findings

Outperforms existing methods on multiple datasets

Captures diverse reaction pathways effectively

Enhances prediction accuracy for rare reaction patterns

Abstract

Reaction prediction, a critical task in synthetic chemistry, is to predict the outcome of a reaction based on given reactants. Generative models like Transformer have typically been employed to predict the reaction product. However, these likelihood-maximization models overlooked the inherent stochastic nature of chemical reactions, such as the multiple ways electrons can be redistributed among atoms during the reaction process. In scenarios where similar reactants could follow different electron redistribution patterns, these models typically predict the most common outcomes, neglecting less frequent but potentially crucial reaction patterns. These overlooked patterns, though rare, can lead to innovative methods for designing synthetic routes and significantly advance synthesis techniques. To address these limitations, we build a team of expert models to capture diverse plausible reaction outcomes for the same reactants, mimicking the divergent thinking of chemists. The proposed framework, ReactionTeam, is composed of specialized expert models, each trained to capture a distinct type of electron redistribution pattern in reaction, and a ranking expert that evaluates and orders the generated predictions. Experimental results across two widely used datasets and different data settings demonstrate that our proposed method achieves significantly better performance compared to existing state-of-the-art approaches.