Retrieval-Augmented Multimodal Learning for Enzyme-Substrate Interaction Prediction Under Low-Homology Shift

2026-06-22

AI Summary

This work addresses the performance degradation in enzyme–substrate interaction (ESI) prediction caused by low sequence homology and extreme sparsity of positive samples. To tackle these challenges, the authors propose RAMMESI, a framework that constructs enzyme–substrate pair representations through directional cross-modal interaction modeling and adaptive fusion. It incorporates a plug-in retrieval-augmented mechanism that, during inference, leverages neighboring enzyme sequences and the query substrate to generate contextual evidence, thereby enhancing robustness. Additionally, an imbalance-aware weighted binary cross-entropy loss is introduced to mitigate data sparsity. Experimental results demonstrate that RAMMESI significantly outperforms existing methods on two benchmark datasets under low-homology splits, and its retrieval module can be generically applied to boost the performance of diverse ESI models.

Abstract

Enzyme substrate interaction (ESI) prediction is a fundamental computational task for biocatalyst discovery and reaction screening in large biochemical spaces. In practical settings, ESI prediction is challenged by sparse positive supervision and low-homology distribution shift, where test enzymes share limited sequence identity with those observed during training. To address these challenges, we propose RAMMESI, a retrieval-augmented multimodal framework for robust ESI prediction. RAMMESI learns explicit pairwise enzyme-substrate representations through directional cross-modal interaction modeling and adaptive fusion. To enhance robustness, RAMMESI retrieves neighboring enzymes at inference time, recombines them with the query substrate, and aggregates the resulting pairwise predictions as contextual evidence. To improve learning under sparse positive supervision, we further adopt an imbalance-aware weighted-BCE objective. Experiments on two ESI benchmarks under sequence-identity-aware splits demonstrate that RAMMESI achieves consistently strong performance, with particular advantages in more challenging low-identity regimes. In addition, the retrieval module improves multiple ESI backbones in a plug-and-play manner, suggesting that retrieval provides a general mechanism for improving robustness under homology shift.

Problem

• enzyme-substrate interaction
• low-homology shift
• sparse positive supervision
• distribution shift
• biocatalyst discovery

Innovation

• retrieval-augmented learning
• multimodal interaction
• enzyme-substrate prediction
• low-homology shift
• imbalance-aware loss