Potential Targets of Atherosclerosis Based on WGCNA and Machine Learning Algorithms
FANG Rourou
ZHANG Hanxiang
YANG Qifan
WU Dongdong
HAN Ruobing
LI Tao
ZHAO Jing
XU Shouzhu
Abstract：Objective:To analyse vascular sequencing dataset of atherosclerosis(AS)using weighted gene co-expression network analysis(WGCNA)and machine learning algorithms to explore potential targets of AS,and providing a basis for clinical treatment and drug development.Methods:The AS gene expression dataset GSE40231 was downloaded from Comprehensive Database of Gene Expression(GEO).Differential expression analysis was conducted on the dataset using the Limma package to obtain differentially expressed genes(DEGs),and WGCNA analysis was performed to determine the gene modules significantly related to AS.The intersection of DEGs and the genes of the modules most significantly related to AS was taken.The intersection genes were identified as the key genes of AS.The key gene-protein interaction network was constructed through the STRING online website,and the key genes were analyzed through gene ontology(GO),Kyoto Encyclopedia of Genes and Genomes(KEGG),and disease ontology(DO).Three machine learning methods(support vector machine,random forest,and minimum absolute shrinkage and selection operator algorithm)were utilized to obtain the Hub gene and analyze its expression.The Hub gene was verified in the datasets GSE20129 and GSE226790.Single-sample gene enrichment(GSEA)was used to further analyze the role and function of the Hub gene in AS diseases.Results:Analysis of the dataset GSE40231 revealed that compared with the arterial samples of healthy controls,a total of 483 DEGs were screened out from the vascular tissue samples of AS,including 245 up-regulated genes and 238 down-regulated genes.Twenty gene modules were constructed through WGCNA.Analysis revealed that eight gene modules were the most significantly related to AS.Enrichment analysis was conducted on the key genes intersecting 8 gene modules in DEGs and WGCNA.GO analysis mainly enriched 138 biological processes(BP),including nuclear cytoplasmic transport,nuclear transport,and organophosphorus catabolic processes.Ten cellular components(CC)included nuclear spots,the apical part of the cell,and the late intracellular part.Forty-three molecular functions(MF)included magnesium ion binding,single-stranded DNA binding and calmodulin binding.KEGG analysis showed significant enrichment in glucagon signaling pathway,tight junction signaling pathway,endocrine and other factors regulating calcium reabsorption signaling pathway,cortisol synthesis and secretion,peroxisomes and other signaling pathway.The glycoprotein hormone subunit α2(GPHA2),lanthanine synthase C-like protein 2(LANCL2.1),C-X9-C motif 4(CMC4),and pre-nuclear mRNA domain 1(RPRD1A.2)were finally screened out as Hub genes by machine learning.Compared with the normal control group,the four Hub genes were significantly downregulated in the AS group,which was consistent with the expression trend in the validation set.Conclusion:GPHA2,LANCL2.1,CMC4,and RPRD11.2 were closely related to the occurrence and development of AS.The four Hub genes might participate in the onset and progression of AS through pathways such as the glucagon signalling pathway.GPHA2,LANCL2.1,CMC4,and RPRD11.2 could serve as novel candidate biomarkers for AS.
Keywords：atherosclerosisweighted gene co-expression network analysismachine learningkey targetbiomarkerbioinformatics
Publication Date：2025-09-25
Online Publishing Date：2025-10-14(First online date of this platform, not the publication date of the document)
Pages：11( 2742-2752 )