■本研究旨在使用生物信息学分析和机器学习算法来识别与骨质疏松症相关的核心基因。
■骨质疏松症患者的mRNA表达谱从基因表达谱(GEO)数据库获得,使用GEO35958和GEO84500作为训练集,和GEO35957和GSE56116作为验证集。使用R软件“limma”软件包进行差异基因表达分析。进行加权基因共表达网络分析(WGCNA)以确定骨质疏松症的关键模块和模块基因。京都基因和基因组百科全书(KEGG),基因本体论(GO),并对差异表达基因进行基因集富集分析(GSEA)。拉索,SVM-RFE,射频机器学习算法被用来筛选核心基因,随后在验证集中进行了验证。还分析了来自核心基因的预测microRNAs(miRNAs),和差异miRNA使用定量实时PCR(qPCR)实验进行验证。
■总共鉴定了1280个差异表达基因。通过WGCNA鉴定了一个疾病关键模块和215个模块关键基因。通过机器学习算法筛选出三个核心基因(ADAMTS5、COL10A1、KIAA0040),COL10A1对骨质疏松有较高的诊断价值。四个核心miRNA(has-miR-148a-3p,has-miR-195-3p,has-miR-148b-3p,has-miR-4531)是通过将预测的miRNA与来自数据集(GSE64433,GSE74209)的差异miRNA相交而发现的。qPCR实验验证了has-miR-195-3p的表达,has-miR-148b-3p,在骨质疏松患者中,has-miR-4531显著升高。
■这项研究证明了生物信息学分析和机器学习算法在识别与骨质疏松症相关的核心基因中的实用性。
UNASSIGNED: This study aimed to identify osteoporosis-related core genes using
bioinformatics analysis and machine learning algorithms.
UNASSIGNED: mRNA expression profiles of osteoporosis patients were obtained from the Gene Expression Profiles (GEO) database, with GEO35958 and GEO84500 used as training sets, and GEO35957 and GSE56116 as validation sets. Differential gene expression analysis was performed using the R software \"limma\" package. A weighted gene co-expression network analysis (WGCNA) was conducted to identify key modules and modular genes of osteoporosis. Kyoto Gene and Genome Encyclopedia (KEGG), Gene Ontology (GO), and gene set enrichment analysis (GSEA) were performed on the differentially expressed genes. LASSO, SVM-RFE, and RF machine learning algorithms were used to screen for core genes, which were subsequently validated in the validation set. Predicted microRNAs (miRNAs) from the core genes were also analyzed, and differential miRNAs were validated using quantitative real-time PCR (qPCR) experiments.
UNASSIGNED: A total of 1280 differentially expressed genes were identified. A disease key module and 215 module key genes were identified by WGCNA. Three core genes (ADAMTS5, COL10A1, KIAA0040) were screened by machine learning algorithms, and COL10A1 had high diagnostic value for osteoporosis. Four core miRNAs (has-miR-148a-3p, has-miR-195-3p, has-miR-148b-3p, has-miR-4531) were found by intersecting predicted miRNAs with differential miRNAs from the dataset (GSE64433, GSE74209). The qPCR experiments validated that the expression of has-miR-195-3p, has-miR-148b-3p, and has-miR-4531 was significantly increased in osteoporosis patients.
UNASSIGNED: This study demonstrated the utility of
bioinformatics analysis and machine learning algorithms in identifying core genes associated with osteoporosis.