基于生物信息学与机器学习破译脊髓损伤中铁死亡相关生物标志物及中药预测  

To decipher ferroptosis-related biomarkers of spinal cord injury and to perform the prediction of traditional Chinese medicine based on bioinformatics and machine learning

作　　者：徐少华 杨晓善 农小霞 张勇 张文捷 Xu Shaohua;Yang Xiaoshan;Nong Xiaoxia;Zhang Yong;Zhang Wenjie(Second Department of Orthopaedics,International Zhuang Medicine Hospital,Guangxi University of Traditional Chinese Medicine,Nanning 530201,China)

机构地区：[1]广西中医药大学附属国际壮医医院骨二科(脊柱外科),广西南宁530201

出　　处：《中国急救医学》2025年第4期306-315,共10页Chinese Journal of Critical Care Medicine

基　　金：广西中西医结合骨与关节退行性疾病多学科交叉创新团队(GZKJ2310)。

摘　　要：目的 探索铁死亡相关基因在脊髓损伤(SCI)中的表达及作用机制,预测SCI铁死亡相关生物标志物,进行药物敏感性检测及中药预测。方法 基于生物信息学与机器学习算法,下载GSE24179数据集,通过limma R包筛选差异基因,构建蛋白互作(PPI)网络进行修剪从而筛选差异表达基因(DEGs)。获得铁死亡相关基因,通过支持向量机(SVM)及受试者工作特征(ROC)曲线进行决策及评估以筛选潜在生物标志物,再进行免疫浸润分析,CellMiner数据库进行药物敏感性预测,LASSO回归模型确定生物标志物。通过TCD数据库及Coremine数据库构建中药-化学成分-基因调控轴,统计性味归经并总结关键中药。结果 获取128个DEGs,生物功能包含脂质代谢、氧化应激与铁死亡密切相关;获得12个铁死亡相关的差异表达基因(FRDEGs),经SVM决策及ROC曲线评估共纳入10个铁死亡相关生物标志物(PFRBs),PFRBs中多数基因存在线性相关,以Arnt样蛋白1(ARNTL)-核受体共激活因子3(NCOA3)相关性最强;免疫浸润分析组间对比显示,ARNTL与NBCs、MBCs、gdTCs、Mono相关皆有统计学意义,与NBCs的相关性尤为显著;药敏预测显示,6-巯基嘌呤对ARNTL及AKR1C3敏感;LASSO回归模型显示,NCOA3、CYP4F8及AKR1C3为保持预测性能的最小变量区间;中药预测结果显示,药味多为苦、甘、辛,药性多寒、温,表里经多为脾胃和肺大肠,生姜、菟丝子、红花、火麻仁是关键中药。结论 ARNT和NCOA3可能通过铁死亡途径介导调控初始B细胞从而干预SCI的免疫微环境,6-巯基嘌呤的药物敏感性最强,部分类药/药性化学成分及中药能够干预SCI,尤其是匹立尼酸及生姜,其中的相关性有待深究。Objective To explore the expression and mechanism of ferroptosis-related genes in spinal cord injury(SCI)based on bioinformatics and machine learning algorithms and to predict potential ferroptosis-related biomarkers(PFRBs)in SCI.To perform drug susceptibility testing and the prediction of traditional Chinese medicine.Methods The GSE24179 dataset was downloaded,the differentially expressed genes(DEGs)were screened by the limma R package,and the protein-protein interaction(PPI)network was constructed for pruning.Ferroptosis-related genes were obtained,and potential biomarkers were screened by support vector machine(SVM)and receiver operating characteristic(ROC)curves for decision-making and evaluation,followed by immune infiltration analysis,drug sensitivity prediction in CellMiner database,and the biomarkers were finally determined by LASSO regression model.Through the TCD database and Coremine database,the traditional Chinese medicine-chemical composition-gene regulation axis was constructed,and the statistical analysis of properties,tastes and meridians was performed and the key traditional Chinese medicines were summarized.Results A total of 128 DEGs were obtained,and their biological functions included lipid metabolism and oxidative stress,which were closely related to ferroptosis.12 ferroptosis-related differentially expressed genes(FRDEGs)were obtained,a total of 10 PFRBs were included through SVM decision-making and ROC curve evaluation,and most genes in PFRBs showed linear correlation,with Arnt-like protein 1(ARNTL)-nuclear receptor coactivator 3(NCOA3)being the most correlated.Immunoinfiltration analysis showed that ARNTL was statistically correlated with NBCs,MBCs,gdTCs and Mono,especially for NBCs.Antimicrobial susceptibility prediction showed that 6-mercaptopurine was sensitive to ARNTL and AKR1C3.The LASSO regression model showed that NCOA3,CYP4F8 and AKR1C3 were the smallest variables to maintain the prediction performance.The prediction results of traditional Chinese medicine showed that the medici

关 键 词：脊髓损伤 铁死亡 生物信息学 机器学习 Arnt样蛋白1(ARNTL) 核受体共激活因子3(NCOA3) 

分 类 号：R28[医药卫生—中药学]