Objective Use bioinformatics and machine learning to find diagnostic biomarkers,pathogenesis,and immune cell infiltration levels that are relevant to people with blood stream infections(BSI)and look for new drug targets.Methods A BSI-related gene expression dataset was obtained from the High-Throughput Gene Expression Omnibus(GEO)database.Use the R language to screen differentially expressed genes(DEG),and then perform gene enrichment analysis.Key module genes were screened using weighted correlation network analysis(WGCNA).Identify central genes by using two machine learning algorithms.Use receiver operating characteristic(ROC)curves and box plot models in external datasets to validate the diagnostic efficacy of central genes.Analyze immune cell infiltration levels using the CIBERSORT deconvolution algorithm.Results This study obtained the intersection genes of 330 weighted gene co-expression network key module genes and DEGs.The results of gene enrichment analysis showed that immune and inflammation-related pathways were significantly enriched.A total of 8 potential biomarkers were obtained through machine learning and external database validation.ROC analysis showed that the area under the curve(AUC)of all 8 potential biomarkers was greater than 0.9.Immunocyte infiltration analysis indicates that all diagnostic biomarkers may have varying degrees of correlation with immune cells.Conclusion Through bioinformatics and machine learning methods,potential biomarkers were identified and a blood flow infection diagnosis model was constructed.This study can provide potential peripheral blood diagnostic biomarkers for patients with bloodstream infections and provide new directions for the pathogenesis of bloodstream infections,new treatment targets,and new drug development.
关键词
血流感染/生物信息学/机器学习/诊断生物标志物/免疫细胞浸润分析
Key words
Bloodstream infection/Bioinformatics/Machine learning/Prediction of biomarkers/Immune cell infiltration
维普
万方数据