摘要
目的利用生物信息学及机器学习方法研究潜在的诊断绝经后骨质疏松症(postmenopausal osteoporosis,PMOP)的生物标志物,为诊断和治疗绝经后骨质疏松症提供新的思路。方法使用GEO数据库下载绝经后骨质疏松症相关的芯片数据集GSE56815和数据集GSE7429,借助R软件Limma包进行数据的差异表达分析,通过R软件ClusterProfiler包对差异基因进行GO功能分析,借助STRING数据库构建调控网络中靶基因的蛋白互作网络,用SVM-RFE分析和LASSO回归模型筛选可能的标志物,利用CIBERSORT软件测定PMOP 22种免疫细胞组成,作关键基因与免疫细胞的相关性分析。结果①数据差异表达分析共获得差异表达基因30个(9个上调,21个下调);②基于蛋白互作网络、LASSO分析、SVM-REF分析得到潜在基因S100A12;③ROC分析得到S100A12的表达具有诊断意义;④S100A12与PMOP中免疫细胞的相关性分析显示显示S100A12与免疫细胞T cells CD8、T cells CD4 memory resting、T cells CD4 memory activated、Plasma cells、Monocytes、Mast cells resting、Macrophages M0、Macrophages M1、Macrophages M2、Eosinophils、Dendritic cells resting等相关。结论S100A12在PMOP中的差异表达具有诊断意义,且可能通过调节多个免疫细胞参与了PMOP的进展,其可能成为潜在的诊断绝经后骨质疏松症的生物学标志物及治疗靶点。
Objective To study the potential biomarkers for the diagnosis of postmenopausal osteoporosis through GEO database,bioinformatics and machine learning method,and provides new ideas for the diagnosis and treatment of postmenopausal osteoporosis.Methods Download the chip dataset GSE56815 and dataset GSE7429 related to postmenopausal osteoporosis through the GEO database,analyzed the differential expression of data with the help of the R software Limma package,and analyzed the differential gene by the R software ClusterProfiler package,built the protein interaction network of the target gene in the regulatory network with the help of the STRING database,and screen for possible markers with SVM-RFE analysis and LASSO regression model.CIBERSORT software was used to determine the composition of PMOP 22 immune cells,and the correlation between key genes and immune cells was analyzed.Results①A total of 30 differentially expressed genes(9 up-regulated and 21 down-regulated)were obtained by differential expression analysis of data.②The potential gene S100A12 was obtained based on LASSO analysis,SVMREF analysis and PPI.③The expression of S100A12 obtained by ROC analysis was diagnostic.Correlation analysis between S100A12 and immune cells in PMOP showed that S100A12 and immune cells T cells CD8 T cells CD4 memory resting T cells CD4 memory activated Plasma cells Monocytes Mast cells rest Macrophages M0 Macrophages M1 Macrophages M2 Eosinophils Dendritic cells rest.Conclusion Based on bioinformatics and machine learning,the key gene S100A12 was screened,and the study showed that the differential expression of S100A12 in PMOP is diagnostic,and may be involved in the progression of PMOP by regulating multiple immune cells,which may become a potential biological marker and therapeutic target for the diagnosis of postmenopausalosteoporosis.
作者
李树栋
梁学振
李刚
LI Shudong;LIANG Xuezhen;LI Gang(The First Clinical Medical College of Shandong University of Traditional Chinese Medicine,Jinan 250014,China)
出处
《中国骨质疏松杂志》
CAS
CSCD
北大核心
2023年第9期1310-1314,共5页
Chinese Journal of Osteoporosis
基金
山东省自然科学基金(ZR2020KH012)
济南市科技计划(202019056)。
