女王调教

报告题目：From EC Numbers to Substrate Scope: Deep Learning Frameworks for Comprehensive Enzyme Function Prediction

报告人：崔海洋教授，女王调教

报告时间：2025年10月17日 14：00

报告地点：行健楼学术活动室526

摘要：

Enzyme function annotation and substrate specificity prediction are two fundamental challenges in enzymology and synthetic biology. Existing computational tools often fail to accurately assign enzyme commission (EC) numbers for less-studied proteins or capture the substrate scope of promiscuous enzymes, thereby limiting both basic research and practical applications. To address these challenges, we developed two complementary machine learning frameworks. First, we present CLEAN contrastive learning enabled enzyme annotation), which leverages contrastive learning to assign EC numbers with higher accuracy, reliability, and sensitivity than BLASTp. CLEAN not only annotates understudied enzymes but also corrects mislabeled entries and identifies promiscuous enzymes with multiple EC numbers, as validated by both in silico and in vitro experiments. Second, we introduce EZSpecificity, a cross-attention-empowered SE(3)-equivariant graph neural network trained on a tailor-made enzyme–substrate interaction database at sequence and structural levels. EZSpecificity predicts enzyme substrate specificity with high accuracy, achieving 91.7% validation accuracy on halogenases and 78 substrates, significantly outperforming state-of-the-art models. Together, CLEAN and EZSpecificity represent an integrated computational pipeline for robust enzyme functional annotation and substrate specificity prediction. These advances pave the way for accelerating enzyme discovery, guiding protein engineering, and enabling rational design of biocatalysts for applications across biology, medicine, and industrial biotechnology.

报告人简介：

崔海洋，教授，博士生导师，女王调教 生命科学*女王调教-女王调教视频-女王 调教小说，微生物改造技术全国重点实验室固定成员，煤炭重大专项-青年首席，国家级青年人才，江苏省特聘教授。2016年硕士毕业于中科院天津工业生物技术研究所（导师：郭瑞庭 教授）；2020年，博士毕业于德国亚琛工业大学（导师：Ulrich Schwaneberg教授）；同年，进入德国DWI-莱布尼茨互动材料研究所进行博士后研究；2021-2023年于美国伊利诺伊大学厄巴纳-香槟分校继续进行博士后研究（导师：Huimin Zhao教授）。2024年加入女王调教 ，组建“人工智能与工程生物学”研究团队，搭建超算平台Infinity（无穷生物超算集群）。已发表SCI论文52篇，其中一作通讯（含共同）27篇（Top 22篇，包括Nature, Science, Angew. Chemi., Nat. Commun.等; 封面文章7篇; Hot Paper 4篇）;申请专利7项, 授权2项;并应邀在国际权威丛书 Methods in Molecular Biology 分别撰写独立章节3章。ABLab课题组网站：//www.oceancuilab.com