摘要
目的 研究机器学习算法鉴定骨质疏松症的诊断标志物。方法 首先从基因表达(GEO)数据库中获取骨质疏松症患者和正常人的GSE56815表达谱数据集。然后使用R软件筛选差异表达基因(DEGs),并进行基因本体(GO)和基因集变异(GSVA)分析。最后使用最小绝对收缩和选择算子(LASSO)逻辑回归及支持向量机递归特征消除(SVM-RFE)2种机器学习算法筛选骨质疏松症的诊断标志物。结果 共得到345个DEGs;通过GO注释研究发现,DEGs与细胞极性和细胞质的转运调节相关;GSVA结果显示,DEGs主要富集在细胞自噬、免疫应答及生物黏附等通路中;共获得了PHF20、DESI2、TRIM44、RAB2A、METTL4、GLT8D2、FCGR2A、ARL4C和EIF3K等9个诊断标志物,其中METTL4和EIF3K可能是骨质疏松症潜在的致病基因。结论 METTL4和EIF3K有望作为骨质疏松症的诊断标志物和治疗靶点。
Objective To study the diagnostic markers of osteoporosis identified by machine learning algo-rithm.Methods Firstly,GSE56815 expression profiles of osteoporosis patients and normal subjects were obtained from the gene expression(GEO)database.Then R software was used to screen differentially expressed genes(DEGs),and gene ontology(GO)and gene set variation(GSVA)analyses were performed.Finally,two machine learning algorithms of least absolute shrinkage and selection operator(LASSO)logistic regression and support vec-tor machine recursive feature elimination(SVM-RFE)were used to screen diagnostic markers of osteoporosis.Re-sults A total of 345 DEGs were obtained.According to GO annotation,DEGs is related to cell polarity and cyto-plasmic transport regulation.The results of GSVA showed that DEGs was mainly concentrated in autophagy,immune response and bioadhesion pathways.A total of 9 diagnostic markers including PHF20,DESI2,TRIM44,RAB2A,METTL4,GLT8D2,FCGR2A,ARL4C and EIF3K were obtained,among which METTL4 and EIF3K may be poten-tial pathogenic genes of osteoporosis.Conclusion METTL4 and EIF3K may be used as diagnostic biomarkers and therapeutic targets for osteoporosis.
作者
马建彪
马永隆
马彦龙
沙如跃
马素俊
唐致军
MA Jianbiao;MA Yonglong;MA Yanlong;SHA Ruyue;MA Sujun;TANG Zhijun(The Second Department of Orthopedics,Guanghe County Integrated Hospital of Traditional Chinese and Western Medicine,Linxia,Gansu,731300,China)
出处
《甘肃中医药大学学报》
2023年第5期40-46,共7页
Journal of Gansu University of Chinese Medicine
关键词
骨质疏松症
差异表达基因
生物信息学
诊断
机器学习
osteoporosis
differentially expressed genes
bioinformatics
diagnosis
machine learning
