Screening of genes associated with chronic sinusitis diagnosis based on machine learning and analysis of their relationship with immune microenvironment
Objective To screen genes related to chronic sinusitis diagnosis by machine learning,and to analyze their correlation with inflammatory cell infiltration in the immune microenvironment of chronic sinusitis.Methods Chronic sinusitis related gene chip was obtained from Gene Expression Omnibus(GEO)public database,and was selected for effective training set;the gene expression difference was analyzed using limma program package of R software.The candidate genes were screened by weighted co-expression analysis,and the chronic sinusitis diagnosis-related gene PLP1 was obtained by three algorithms.ROC curve was used to evaluate the diagnostic value of PLP1 in chronic sinusitis,CIBERSORT algorithm was used to analyze the immune microenvironment of chronic sinusitis,and the correlation between PLP1 and immune cell infiltration was analyzed.Results A total of 184 related genes were detected while comparing chronic sinusitis samples and normal control samples.These related genes involved in biological processes such as white blood cell migration and cell activation of immune response.Three algorithms were used to screen related genes,and finally PLP1 was found to be the relevant gene for chronic sinusitis diagnosis.PLP1 gene was down-regulated in both training set(GSE23552)and validation set(GSE179265),and the difference was statistically significant(P＜0.05).The ROC curve showed that the AUC of PLP1 was 1.000 in the training set while 0.950 in the verification set.PLP1 gene expression was correlated with a variety of immune cell infiltration,and the correlation coefficient(r)between PLP1 expression and eosinophilic cell infiltration was-0.7(P＜0.001).Conclusion PLP1 can be used for diagnosis of chronic sinusitis,whose expression level is negatively correlated with eosinophils and M2 macrophage infiltration.
万方数据
维普
