摘要
木质纤维素是植物细胞壁的主要成分,是地球上最丰富且可再生的生物质资源,其富含多糖,具有开发为家畜饲料资源的潜力巨大。但木质纤维素紧密复杂的结构限制了家畜对其的消化利用。利用微生物及其分泌的酶系降解木质纤维素,具有低能耗、少污染、高环境兼容性等优势。通过深度学习和机器学习等人工智能技术,可高效地从酶序列和酶结构中提取关键特征并进行性能预测,能够有效解决传统酶工程实验费时、耗力的问题。本文综述了人工智能在木质纤维素生物降解领域中酶工程应用方面的研究进展,以期为木质纤维素在饲料资源开发及绿色生物制造领域的研究与应用提供参考。
Lignocellulose,the primary component of plant cell walls,is the most abundant and renewable biomass resource on earth.Its rich polysaccharide content holds tremendous potential for the development of livestock feed resources.However,the intricate and recalcitrant structure of lignocellulose limits its digestibility and utilization by livestock.Biological degradation of lignocellulose using microorganisms and their secreted enzymes offers advantages including low energy consumption,minimal pollution,and high environmental compatibility.By employing artificial intelligence algorithms such as deep learning and machine learning,researchers can efficiently extract key features from enzyme sequences and structures while predicting their performance.These approaches can effectively address the time-consuming and labor-intensive challenges associated with traditional enzyme engineering.This review summarized recent advances in applying artificial intelligence to lignocellulase engineering for lignocellulose biodegradation,aiming to provide insights for the development of lignocellulose-based feed resources and sustainable green manufacturing.
作者
周家安
高乐
李忠秋
刘春龙
ZHOU Jiaan;GAO Le;LI Zhongqiu;LIU Chunlong(Northeast Institute of Geography and Agroecology,Chinese Academy of Sciences,Harbin 150081,China;University of Chinese Academy of Sciences,Beijing 100049,China;Tianjin Institute of IndustrialBiotechnology,Chinese Academy of Sciences,Tianjin 300308,China;Institute of Animal Science,Heilongjiang Academy of Agricultural Sciences,Harbin 150086,China)
出处
《动物营养学报》
北大核心
2025年第9期5811-5823,共13页
CHINESE JOURNAL OF ANIMAL NUTRITION
基金
中国科学院战略性先导科技专项(XDC0110304)
黑龙江省重点研发计划(创新基地)项目(JD24A012)。
关键词
人工智能
木质纤维素
酶工程
深度学习
机器学习
artificial intelligence
lignocellulase
enzyme engineering
deep learning
machine learning
