机器学习助力酶定向进化  

Machine learning-assisted enzyme directed evolution

作　　者：蒋迎迎 曲戈 孙周通 JIANG Ying-ying;QU Ge;SUN Zhou-tong(Tianjin Institute of Industrial Biotechnology,Chinese Academy of Sciences,Tianjin 300308,China)

机构地区：[1]中国科学院天津工业生物技术研究所,天津300308

出　　处：《生物学杂志》2020年第4期1-11,共11页Journal of Biology

基　　金：国家重点专项(2019YFA0905100);国家自然科学基金(No.31870779,31900909);天津市自然科学基金(No.18JCYBJC24600,19JCQNJC09100)。

摘　　要：酶定向进化技术在生物催化、生物医药、生物技术等领域扮演重要角色。得益于计算速度的大幅提升以及海量数据集的出现,当前人工智能技术发展如火如荼。近年来机器学习等人工智能方法也被应用于蛋白质工程,在复杂酶结构预测、稳定性/选择性/可溶性、指导酶设计等问题中表现出独特的优势,为酶分子设计提供了新的可能。综述了当前机器学习算法及描述符助力酶设计改造方面的应用与进展。Directed evolution plays a central role in the fields of biocatalysis,biomedicine and biotechnology,etc.Taking advantages of increasingly computer performance and numerous datasets,artificial intelligence has rapidly developed.Recently,machine learning algorithms have also been applied to protein engineering,especially in helping prediction of protein structures,improving enzyme stability/selectivity/solubility,and guiding rational protein design as well as other functions.This paper reviews the state of the art in algorithms and descriptors used in enzyme engineering.

关 键 词：人工智能 蛋白质工程 定向进化 机器学习 

分 类 号：Q503[生物学—生物化学] Q814