基于生物信息学和机器学习筛选扩张型心肌病核心基因

Screening of Hub Genes for Dilated Cardiomyopathy Based on Bioinformatics and Machine Learning

目的:通过生物信息学和机器学习筛选扩张型心肌病(DCM)核心基因。方法:从基因表达数据库中获取GSE42955和GSE84796芯片数据集。先用R软件“limma”包初筛差异表达基因(DEGs),然后通过构建加权基因共表达网络(WGCNA)来确定DCM关键模块和模块核心基因;再通过最小绝对值收敛和选择算子(LASSO)、随机森林两种机器学习算法筛选疾病核心基因;最后对核心基因进行基因本体论(GO)和京都基因与基因组百科全书(KEGG)通路富集分析,并基于上述数据集绘制核心基因受试者工作特征(ROC)曲线以评估核心基因诊断DCM的价值。结果:在GSE42955和GSE84796数据集中,共鉴定出179个DEGs (139个上调和40个下调);通过WGCNA获得1个疾病关键模块及99个模块关键基因;通过机器学习最终筛选出3个DCM核心基因杀伤细胞凝集素样受体C3(KLRC3)、淋巴细胞跨膜衔接子1(LAX1)、脊髓灰质炎病毒受体相关免疫蛋白结构域(PVRIG),它们主要富集于免疫应答。核心基因ROC显示3个核心基因对DCM均具有较高诊断价值,曲线下面积(AUC)分别为0.962、0.962和0.936。结论:KLRC3、LAX1、PVRIG为DCM的核心基因,或可成为诊断DCM的潜在生物标志物。

Objective: To screen the hub genes of dilated cardiomyopathy(DCM) by bioinformatics and machine learning. Method: The microarray datasets of GSE42955 and GSE84796 were obtained from the gene expression database. Firstly, the “limma” package of R software was applied to screen the differentially expressed genes(DEGs), then a weighted gene co-expression network(WGCNA) was constructed to identify the hub modules and module hub genes of DCM. Then the disease hub genes were screened by two machine learning algorithms, Least Absolute Shrinkage and Selection Operator(LASSO) and Random Forest. Finally, the hub genes were subjected to gene ontology(GO) and kyoto Encyclopedia of genes and genomes(KEGG) pathway enrichment analyses, and hub genes receiver operating characteristic(ROC) curve were plotted based on the above datasets to assess the value of hub genes in diagnosing DCM. Results: In the GSE42955 and GSE84796 datasets, 179 DEGs(139 up-regulated and 40 down-regulated DEGs) were identified;1 disease key module and 99 module hub genes were obtained by WGCNA;3 DCM hub genes(KLRC3,LAX1,PVRIG) were finally screened by machine learning, and the hub genes were mainly enriched in immune response.The ROC curve of the hub genes showed that all 3 hub genes had high diagnostic value for DCM, and the area under curve(AUC) was 0.962, 0.962 and 0.936.Conclusion: KLRC3, LAX1 and PVRIG are the hub genes of DCM, which may be potential biomarkers for the diagnosis of DCM.