Search papers, labs, and topics across Lattice.
The paper introduces RxnCLF, a self-supervised contrastive learning framework that utilizes a condensed reaction graph (CRG) to enhance reaction yield prediction by integrating reactant and product information into a unified structure. By pretraining on a dataset of 1.7 million reactions, RxnCLF learns a compact latent space that captures both reaction-center features and broader side chain contexts, leading to improved chemical interpretability. Fine-tuning on various yield prediction benchmarks demonstrates that RxnCLF consistently outperforms existing graph and sequence-based models, achieving superior performance across multiple tasks.
RxnCLF outperforms traditional models by leveraging a unified reaction graph, revealing a new paradigm for reaction yield prediction that enhances interpretability and generalization.
Reaction yield prediction remains challenging because labeled data are scarce and reaction space is both combinatorially large and sparsely populated, limiting the generalization of existing reaction representations. String-, fingerprint-, and graph-based reaction encodings only partially capture chemical transformations, making accurate prediction difficult for reactions with complex substrates. We propose reaction contrastive learning foundation (RxnCLF), a self-supervised contrastive framework for reaction representation learning. RxnCLF is built on a condensed reaction graph (CRG) that unifies reactant and product information into a single graph, enabling the model to learn explicit and enriched transformation structure rather than disconnected graphs. Pretrained on 1.7 million Pistachio reactions, RxnCLF learns a compact and continuous latent space that captures both reaction-center features and broader side chain contexts, making it transformation-aware and chemically interpretable. Fine-tuned on multiple yield prediction benchmarks, including Buchwald-Hartwig, Pd-catalyzed BH coupling, and proprietary HTE C-N coupling and amide formation datasets, RxnCLF consistently outperforms graph and sequence-based baselines, improving R2 and achieving the best performance overall. Our results highlight the promise of CRG-based RxnCLF as a scalable reaction foundation model, with the potential to generalize across broader reaction spaces and support diverse downstream reaction informatics tasks, including regioselectivity prediction, enantioselectivity prediction, and reaction condition optimization.
