Enzypick

EnzyPick is an innovative web server for high-throughput enzyme screening powered by the deep learning model SPEPP (Substrate-product Pair-based Enzyme Promiscuity Prediction). It enables researchers to rapidly identify enzymes capable of catalyzing specific biochemical transformations between defined substrate-product pairs, bypassing traditional reliance on Enzyme Commission (EC) numbers and prior reaction knowledge.

EnzyPick uniquely supports user-defined custom enzyme libraries, allowing screening against any protein sequence set. Its SPEPP model utilizes transformer architecture and transfer learning to predict enzyme promiscuity scores, indicating the catalytic likelihood for a given pair. The platform provides intuitive visualizations, including atom-to-atom mapping of reference reactions and enzyme structure views highlighting attention weights critical for catalysis, facilitating downstream enzyme engineering.

Designed for accessibility, EnzyPick empowers metabolic engineers and synthetic biologists—even without programming expertise—to streamline enzyme discovery for diverse tasks. Applications include designing novel biosynthetic pathways, optimizing biocatalytic processes, discovering enzymes for hazardous material degradation, and exploring alternative substrates/products for known enzymes. EnzyPick offers a powerful, flexible solution for accelerating enzyme-driven innovation.