基于机器学习筛选类风湿关节炎的诊断标志基因和免疫浸润分析

Machine learning-based screening of diagnostic marker genes for rheumatoid arthritis and analysis of immune infiltration

目的基于机器学习筛选类风湿关节炎(RA)的诊断标志基因,并分析可能的免疫浸润机制,为RA的临床治疗提供参考。方法从基因表达综合(GEO)数据库下载RA基因表达谱芯片数据集,将GSE55235和GSE77298作为联合芯片训练集,GSE55457作为独立验证数据集。使用R软件进行差异表达基因(DEGs)的筛选,并对这些DEGs进行基因本体论(GO)富集分析及京都基因与基因组百科全书(KEGG)富集分析。进一步应用三种机器学习算法筛选诊断基因,并进行外部验证和受试者工作特征(ROC)曲线分析。通过xCell算法分析免疫细胞在RA中的浸润情况。结果筛选出RA的DEGs共704个。富集分析发现这些DEGs主要涉及白细胞介导的免疫、免疫应答的激活、白细胞迁移等相关免疫功能,以及趋化因子信号通路、利什曼病、类风湿关节炎等相关炎症通路。通过机器学习筛选出4个诊断基因,包括趋化因子CXC配体13(CXCL13)、富含亮氨酸重复序列结构域15(LRRC15)、多配体蛋白聚糖-1(SDC-1)和核酸结合蛋白3(YBX3)。免疫浸润分析结果显示,在RA中B细胞、CD4^(+)T细胞、树突状细胞和单核细胞的水平显著上调(P<0.05)。结论RA的发生发展是多基因、多通路共同参与的结果,CXCL13、LRRC15、SDC-1和YBX3可能是诊断RA的潜在生物标志物。B细胞、CD4^(+)T细胞、树突状细胞和单核细胞可能在RA的发生中具有重要意义。

Objective To screen the diagnostic marker genes of rheumatoid arthritis(RA)and analyze the possible immune infiltration mechanism based on bioinformatics and machine learning,and to provide reference for the clinical treatment of RA.Methods The gene expression profiles were downloaded from the Gene Expression Omnibus(GEO)database.GSE55235 and GSE77298 were used as the combined chip training set,and GSE55457 was used as the independent validation dataset.The differentially expressed genes(DEGs)were screened using R software,and Gene Ontology(GO)enrichment analysis and Kyoto Encyclopedia of Genes and Genomes(KEGG)enrichment analysis were performed for these DEGs.Three machine learning algorithms were further applied to screen the diagnostic genes and perform the external validation and receiver operating characteristic(ROC)curve analysis.Finally,the xCell method was used to calculate the infiltration of immune cell types in the RA.The infiltration of immune cells in RA was analyzed by using xCell algorithm.Results A total of 704 DEGs of RA were screened.The results of enrichment analysis revealed that these DEGs were mainly involved in some related immune functions,such as leukocyte-mediated immunity,activation of immune response,and leukocyte migration,and some inflammatory pathways,such as chemokine signaling pathway,Leishmaniasis and Rheumatoid arthritis.Four diagnostic genes,including C-X-C motif chemokine ligand 13(CXCL13),leucine rich repeat containing 15(LRRC15),syndecan 1(SDC-1)and Y-box binding protein 3(YBX3),were screened using machine learning.The results of the immune infiltration analysis showed that the expression levels of B cells,CD4^(+)T cells,dendritic cells and monocytes were significantly up-regulated in RA.Conclusion Multiple genes and pathways are involved in the occurrence and development of RA.CXCL13,LRRC15,SDC-1 and YBX3 may be the potential biomarkers for the diagnosis of RA.Moreover,B cells,CD4^(+)T cells,dendritic cells and monocytes may play an important role in the occurrence of RA.