【摘要】 目的：利用生物信息学方法筛选牙周炎症组织中5-甲基胞嘧啶（m5C）相关生物标志物，并进一步探究其在牙周炎诊断及分型中的应用价值。方法：在GEO数据库中获取牙周炎数据集，对牙周炎样本与正常对照样本进行差异分析，将差异基因与m5C调节因子取交集获得生物标志物。构建随机森林模型并绘制ROC曲线以评估生物标志物的诊断性能。基于生物标志物对牙周炎样本进行分型。最后，在外部数据集中进行验证。结果：牙周炎样本与健康对照样本有1107个差异基因，与m5C调节因子交叉后获得包括TET2和UNG在内的2个生物标志物，它们在牙周炎症组织中均呈下降表达，随机森林模型的混淆矩阵及ROC曲线下面积证明了生物标志物的诊断价值。一致性聚类分析表明，聚类为两个亚型最为合适。结论：本研究确定了2个m5C相关生物标志物（TET2和UNG）与牙周炎相关，它们在牙周炎的诊断及分型中发挥重要作用。本研究为牙周炎的生物信息学诊断及治疗提供了新的参考。更多还原

【Abstract】 Objective:To screen 5-methylcytosine(m5C)-related biomarkers in periodontitis tissues by bioinformatics methods,and to further explore their application value in the diagnosis and classification of periodontitis.Methods:The periodontitis data set was obtained from the GEO database,and the differences between periodontitis samples and normal control samples were analyzed.The differential genes and m5C regulators were intersected to obtain biomarkers.Random forest model was constructed and ROC curve was drawn to evaluate the diagnostic performance of biomarkers.The periodontitis samples were classified based on biomarkers,and the immune microenvironment of different subtypes of periodontitis was further analyzed.Finally,validation was performed in an external dataset.Results:A total of 1 107 differentially expressed genes were identified between the periodontitis samples and the healthy controls.Two biomarkers,including TET2 and UNG,were obtained after cross-validation with m5C regulators,and their expressions were down-regulated in the inflammatory tissues.The confusion matrix and the area under the ROC curve of the random forest model demonstrate the diagnostic value of the biomarkers.Consensus clustering analysis showed that two subtypes were most appropriate.Conclusion:This study identified two m5C-related biomarkers(TET2 and UNG) that were associated with periodontitis and played important roles in the diagnosis and classification of periodontitis.This study provides a new reference for the bioinformatics diagnosis and treatment of periodontitis.更多还原