AI Insight
Researchers developed EzHit, a computational framework that matches enzymes to the chemical reactions they catalyze, enabling systematic discovery of biocatalysts from large metagenomic databases. The system uses geometric confidence measures to prioritize reliable predictions and was applied to screen nearly one trillion protein sequences, identifying 8.20 billion enzyme-reaction associations across 30.27 million predicted enzymes. Experimental validation confirmed that EzHit successfully identified active enzymes for a poorly characterized biochemical reaction.
Why it matters
This tool addresses a major bottleneck in biotechnology by efficiently connecting uncharacterized enzymes to their functions, which could accelerate the discovery of biocatalysts for sustainable chemical synthesis, drug production, and industrial applications. The scalability to metagenomic datasets opens access to the vast diversity of microbial enzymes that remain unexplored.
Understand the Science
⚠️ Preprint – Noch nicht peer-reviewed
Dieser Artikel wurde noch nicht von unabhängigen Experten begutachtet. Die Ergebnisse sind vorläufig und sollten mit Vorsicht interpretiert werden.
Enzyme catalyzed reactions drive metabolism and enable diverse applications in biotechnology and sustainable synthesis, but assigning catalytic functions to uncharacterized enzymes and finding enzymes for target reactions remain major challenges. Despite recent advances in machine learning, translating enzyme function prediction into systematic discovery across the vast space of uncharacterized proteins and biochemical reactions remains challenging. Here we present EzHit, a lightweight framework for bidirectional enzyme reaction retrieval. Across enzyme annotation, orphan-reaction retrieval and family-specific fine-tuning benchmarks, EzHit achieved strong top-ranked performance and outperformed representative enzyme reaction matching baselines. We further show that Mahalanobis distance based geometric confidence provides more useful reliability estimates than probability-derived uncertainty metrics or evidential learning, enabling confidence-aware prioritization of high scoring candidates. Owing to its efficiency, EzHit was applied to a nearly trillion-scale OMG_Prot50 with RHEA screen, yielding 8.20 billion retained enzyme reaction associations across 30.27 million predicted enzyme sequences. Experimental validation further confirmed the activity of EzHit prioritized metagenomic enzyme candidates for a sparsely annotated reaction. Together, these results establish EzHit as a scalable framework for enzyme annotation, reaction-guided mining and biocatalyst discovery
Source: Scalable Enzyme Reaction Retrieval Enables Metagenomic Biocatalyst Discovery