特发性肺纤维化中基底膜相关标志物探索及治疗药物预测

Exploration of Basement Membrane-related Markersand Prediction of Therapeutic Drugs in Idiopathic Pulmonary Fibrosis

目的探索特发性肺纤维化(IPF)中基底膜标志物及潜在治疗药物。方法在基因表达综合数据库(GEO)下载IPF相关数据集,处理后构建与IPF相关的基底膜基因表达矩阵并筛选基底膜差异基因(DEBMs);将DEBMs进行功能及通路的富集,并对其使用机器学习算法得到候选特征基因,运用接收机工作特性(ROC)曲线确定特征基因并构建列线图;进行ssGSEA分析探究特征基因与免疫细胞及功能相关性;通过特征基因预测了相应微小核糖核酸(RNA)(miRNA)及治疗药物。结果共提取DEBMs 56个;富集分析表明,DEBMs主要富集在“细胞外基质组织”、“细胞外结构组织”等,并与“ECM-受体相互作用”和“局部粘着斑”等通路密切相关,机器学习计算到候选特征基因6个(TIMP3、P3H2、ITGA7、ITGA4、ADAMTS2、COL8A2),经ROC曲线测试均符合特征基因要求,列线图诊断价值突出(AUC=0.991523);IPF中B cells、Macrophages等与正常组有显著差异。最后,预测到miRNA以miR-4305、miR-3684为主,黄体酮,叔丁基过氧化氢等是与IPF相关性较强的治疗药物。结论特征基因及预测的miRNA可作为IPF诊断新型标志物,预测药物可能成为治疗IPF的潜在药物来源。

Objective To explore basement membrane markers and potential drugs for treatment in idiopathic pulmonary fibrosis(IPF).Methods IPF-related datasets were downloaded from the Gene Expression Omnibus(GEO)database,processed to construct basement membrane gene expression matrices associated with IPF,and screened for differential basement membrane genes(DEBMs);DEBMs were enriched for function and pathways,and machine learning algorithms were used to obtain candidate signature genes,receiver operating characteristic(ROC)curves were used to identify signature genes and construct a nomogram.We performed ssGSEA analysis to explore the correlation between signature genes and immune cells and their functions and predicted the corresponding miRNAs and therapeutic drugs by signature genes.Results A total of 56 DEBMs were extracted;enrichment analysis showed that DEBMs were mainly enriched in"extracellular matrix tissue","extracellular structural tissue",etc.,and were closely related to"ECM-receptor interaction"and"local adhesion spot"pathways.The machine learning has identified six candidate signature genes(TIMP3,P3H2,ITGA7,ITGA4,ADAMTS2,COL8A2),all of which meet the requirements of the signature genes by the ROC curve test,and the nomogram diagnostic value was outstanding(AUC=0.991523);B cells and Macrophages in IPF were significantly different from the normal group.Finally,miRNAs were predicted to be dominated by miR-4305,miR-3684,progesterone,and tert-butyl hydroperoxide as therapeutic agents with strong relevance to IPF.Conclusion Signature genes and predictive miRNAs may serve as novel markers for IPF diagnosis,and predictive drugs may be a potential source of drugs for treating IPF.