Learning Continuous Solvent Effects from Transient Flow Data: A Graph Neural Network Benchmark on Catechol Rearrangement

TL;DR

This work tackles the problem of predicting reaction yields across continuously varying solvent compositions in transient flow data. It introduces the Catechol Benchmark and demonstrates that a hybrid Graph Neural Network, combining molecular graph attention, precomputed differential reaction fingerprints, and learned mixture encodings, achieves dramatically superior accuracy (MSE ≈ 0.0039) compared with traditional tabular methods and large language models. Ablation studies show explicit molecular structure, kinetic signatures, and non-additive mixture modeling are all essential for robust generalization to unseen solvents and solvent mixtures. The results advocate for structure-centered representations in chemical predictive modeling and provide open-source data and code to accelerate progress in data-efficient, continuous-condition chemistry modeling.

Abstract

Predicting reaction outcomes across continuous solvent composition ranges remains a critical challenge in organic synthesis and process chemistry. Traditional machine learning approaches often treat solvent identity as a discrete categorical variable, which prevents systematic interpolation and extrapolation across the solvent space. This work introduces the \textbf{Catechol Benchmark}, a high-throughput transient flow chemistry dataset comprising 1,227 experimental yield measurements for the rearrangement of allyl-substituted catechol in 24 pure solvents and their binary mixtures, parameterized by continuous volume fractions ($\% B$). We evaluate various architectures under rigorous leave-one-solvent-out and leave-one-mixture-out protocols to test generalization to unseen chemical environments. Our results demonstrate that classical tabular methods (e.g., Gradient-Boosted Decision Trees) and large language model embeddings (e.g., Qwen-7B) struggle with quantitative precision, yielding Mean Squared Errors (MSE) of 0.099 and 0.129, respectively. In contrast, we propose a hybrid GNN-based architecture that integrates Graph Attention Networks (GATs) with Differential Reaction Fingerprints (DRFP) and learned mixture-aware solvent encodings. This approach achieves an \textbf{MSE of 0.0039} ($\pm$ 0.0003), representing a 60\% error reduction over competitive baselines and a $>25\times$ improvement over tabular ensembles. Ablation studies confirm that explicit molecular graph message-passing and continuous mixture encoding are essential for robust generalization. The complete dataset, evaluation protocols, and reference implementations are released to facilitate data-efficient reaction prediction and continuous solvent representation learning.

Figures (6)

• Figure 1: The figure Above shows the architecture of GNN model.
• Figure 2: Above figure shows the Molecular Graph representation, each molecule(starting material, products, and solvents) is represented as a graph.
• Figure 3: Above figure shows the loss of the DeepModel in training. It can be seen that the model performs well in both stages overall, but it performs even better in the mixed solvent.
• Figure 4: The figure above illustrates the convergence profiles of the GNN model under two distinct experimental settings.
• Figure 5: The comparison of single / mixture solvent of GBDT + DeepModel +Ensemble