植物基因组语言模型的研究进展与应用OA
Plant Genomic Language Models:Advances and Applications
大语言模型(LLMs)的发展,为基因组学研究提供了新路径.基因组语言模型(gLMs)的核心思想是将基因组数据视为具有内在语法的生物语言,对其建模从而挖掘隐含的调控逻辑,为解析遗传机制提供了新方法.该文围绕植物基因组语言模型的理论基础与研究现状展开综述,介绍了基于跨物种预训练模型预测变异效应与功能元件、面向单细胞数据的细胞类型注释以及多模态数据融合3个重要研究方向,讨论了植物基因组中高度重复序列和多倍体等特征对模型构建的影响,阐述了部分已有建模策略与模型部署框架,并展望了植物基因组语言模型的发展前景.
The development of large language models(LLMs)has opened new avenues for genomics research,leading to the emergence of genomic language models(gLMs).gLMs treat genomic data as a biological language with intrinsic grammar,learning the regulatory logic embedded within the genome and offering new approaches for the systematic dissection of genetic regulatory mechanisms.This review focuses on the theoretical foundations and current research landscape of plant genomic language models,and discusses major research directions:the application of cross-species pretrained models to variant effect prediction and functional element annotation,the development of models tailored to single-cell data,and strategies for integrating multimodal genomic information.We further discuss how plant-specific genomic features,particularly repetitive elements and polyploidy,affect model construction,and review modeling strate-gies and deployment frameworks.We conclude with a perspective on the future development of plant genomic language models.
张枭;李琼;王楠
中国热带农业科学院热带作物品种资源研究所,海口 571101中国热带农业科学院热带作物品种资源研究所,海口 571101中国农业科学院农业基因组研究所,岭南现代农业科学与技术广东省实验室深圳分中心,热带作物生物育种全国重点实验室,农业农村部农业基因数据分析重点实验室,深圳 518124
植物基因组学人工智能大语言模型多组学遗传变异
plant genomicsartificial intelligencelarge language modelsmulti-omicsgenetic variation
《植物学报》 2026 (4)
571-587,17
国家自然科学基金(No.32302498)
评论