• 中国核心期刊数据库收录期刊
  • 中文科技期刊数据库收录期刊
  • 中国期刊全文数据库收录期刊
  • 中国学术期刊综合评价数据库统计源期刊等

快速检索引用检索图表检索高级检索

中国医药导刊 ›› 2025, Vol. 27 ›› Issue (9): 919-929.

• 医药信息学 • 上一篇    下一篇

基于生物信息学及机器学习探讨苍术治疗非酒精性脂肪肝的作用机制

黄馨妮1, 卢艾华2, 张舒2, 方汝雪3, 刘旭凌1, 李俊雄4*   

  1. 1.上海市普陀区中心医院,上海 200333;

    2.上海市真新社区卫生服务中心,上海 201824;

    3.昆山市第四人民医院,江苏 昆山 215300;

    4.复旦大学附属华东医院,上海 200040

  • 收稿日期:2025-03-18 修回日期:2025-09-02 接受日期:2025-11-12 出版日期:2025-09-28 发布日期:2025-11-18
  • 基金资助:

    上海市普陀区科委项目(ptkwws202223)

Integrating Bioinformatics and Machine Learning Algorithms to Elucidate Target Characteristics and Molecular Mechanisms of CangZhu in Non-Alcoholic Fatty Liver Disease Intervention

HUANG Xinni1, LU Aihua2, ZHANG Shu2, FANG Ruxve3, LIU Xuling1, LI Junxiong4*   

  1. 1.Shanghai Putuo District Central Hospital Shanghai 200333, China
    2.Shanghai Zhenxin Community Health Service Center Shanghai 201824, China
    3.Kunshan Fourth People's Hospital Jiangsu Kunshan 215300, China
    4.Huadong Hospital Fudan University Shanghai 200040, China
  • Received:2025-03-18 Revised:2025-09-02 Accepted:2025-11-12 Online:2025-09-28 Published:2025-11-18

摘要:

目的:旨在运用网络药理学结合生物信息学方法探讨苍术治疗非酒精性脂肪性肝病( non-alcoholic fatty liver disease NAFLD)的潜在分子机制。方法:采用网络药理学方法获取苍术的潜在的靶点,通过检索基因表达数据库(gene expression omnibusGEO)获取NAFLD的靶点,并通过构建靶点PPI网络筛选核心目标靶点。运用R4.2.2软件对靶点进行差异分析、相关性分析,以获得显著差异表达的核心基因(SDECGs)。接着,对SDECGs进行富集分析和免疫浸润分析。利用SDECGs构建机器学习模型以筛选特征基因并构建Nomo图。采用分子对接和GSE63067数据集外部验证法对上述结果进行初步验证。结果:本研究确定了NAFLD和正常组之间的14SDECGs,这些SDECGs主要富集于IL-17AGE-RAGE等信号通路,且与免疫细胞存在多种相互调节关系。XGBoost模型(XGB)是多种机器学习中最优的模型,确定了CTNNB1IL10PTGS2IL6JUNXGB模型中前5个特征基因,并用于构建nomo图。外部数据集证明了该模型的可靠性(AUC值为0.73)。最后,分子对接确定了苍术核心有效成分(OB值排名前4位)(MOL000179MOL000186MOL000449以及MOL000188)与NAFLD特征基因之间可形成稳定结构。结论:苍术及其有效活性成分可能通过作用CTNNB1IL10PTGS2IL6JUN,并调节免疫和炎症相关途径从而缓解NAFLD,这些发现对临床实践和未来的研究具有潜在影响。


关键词: 非酒精性脂肪性肝病, 苍术, 机器学习, 网络药理学, 分子对接

Abstract:

Objective To explore the potential molecular mechanism of CangZhu in the treatment of non-alcoholic fatty liver disease NAFLD by network pharmacology combined with bioinformatics methods. Methods Network pharmacology method was used to obtain the potential targets of CangZhu. The targets of NAFLD were obtained by searching the Gene Expression Omnibus GEO database and the core targets were screened by constructing the target PPI network. Then R4.2.2 software was used for differential analysis and correlation analysis of the target genes to obtain the core genes with significant differential expression SDECGs. Next enrichment analysis and immune infiltration analysis of SDECGs were performed. SDECGs was used to construct a machine learning model to screen feature genes and construct a Nomo map. Finally the molecular docking and external validation methods of GSE63067 dataset were used to validate the above results. Results Fourteen SDECGs were identified between NAFLD and normal subjects. These SDECGs were mainly enriched in IL-17 AGE-RAGE and other signaling pathways and had a variety of mutual regulatory relationships with immune cells. XGBoost model XGB is the best model in a variety of machine learning. CTNNB1 IL10 PTGS2 IL6 and JUN were identified as the top five feature genes in the XGB model and used to construct the nomo map. External data sets demonstrated the reliability of the model AUC=0.73. Finally molecular docking confirmed that the core active components of CangZhu MOL000179 MOL000186 MOL000449 and MOL000188 could form a stable structure with NAFLD characteristic genes. Conclusion CangZhu and its active components may alleviate NAFLD by regulating CTNNB1 IL10 PTGS2 IL6 JUN and immune-and inflammation-related pathways. These findings have potential implications for clinical practice and future research.

 

Key words: Nonalcoholic fatty liver disease , Atractylodis Rhizoma , Machine learning , Network pharmacology , Molecular docking

中图分类号: