Co-expression network analysis identifies potential candidate hub genes in severe influenza patients needing invasive mechanical ventilation

Abstract

Background: Influenza is a contagious disease that affects people of all ages and is linked to considerable mortality during epidemics and occasional outbreaks. Moreover, effective immunological biomarkers are needed for elucidating aetiology and preventing and treating severe influenza. Herein, we aimed to evaluate the key genes linked with the disease severity in influenza patients needing invasive mechanical ventilation (IMV). Three gene microarray data sets (GSE101702, GSE21802, and GSE111368) from blood samples of influenza patients were made available by the Gene Expression Omnibus (GEO) database. The GSE101702 and GSE21802 data sets were combined to create the training set. Hub indicators for IMV patients with severe influenza were determined using differential expression analysis and Weighted correlation network analysis (WGCNA) from the training set. The receiver operating characteristic curve (ROC) was also used to evaluate the hub genes from the test set's diagnostic accuracy. Different immune cells' infiltration levels in the expression profile and their correlation with hub gene markers were examined using single-sample gene set enrichment analysis (ssGSEA).

Results: In the present study, we evaluated a total of 447 differential genes. WGCNA identified eight co-expression modules, with the red module having the strongest correlation with IMV patients. Differential genes were combined to obtain 3 hub genes (HLA-DPA1, HLA-DRB3, and CECR1). The identified genes were investigated as potential indicators for patients with severe influenza who required IMV using the least absolute shrinkage and selection operator (LASSO) approach. The ROC showed the diagnostic value of the three hub genes in determining the severity of influenza. Using ssGSEA, it has been revealed that the expression of key genes was negatively correlated with neutrophil activation and positively associated with adaptive cellular immune response.

Conclusion: We evaluated three novel hub genes that could be linked to the immunopathological mechanism of severe influenza patients who require IMV treatment and could be used as potential biomarkers for severe influenza prevention and treatment.

Keywords: Co-expression network analysis; Hub gene; Invasive mechanical ventilation; Severe influenza.

Fig. 1

Illustrates the training set gene expression profiling. A A heatmap of the top 50 DEGs. Upregulated genes are seen in red, while downregulated genes are highlighted in blue. B The DEGs volcanic plot. Upregulated genes are highlighted in red, while downregulated genes are indicated in green

Fig. 2

The DEGs from the training set were analyzed for functional enrichment Analysis of; A GO enrichment, and B Analysis of KEGG enrichment

Fig. 3

Establishing a weighted gene co-expression network analysis (WGCNA) and screening for Hub genes. A Scale-free fit index and mean connectivity analysis for various soft-thresholding powers. The red line denotes the point when the correlation coefficient is 0.9, and the soft-thresholding power (β) is 6. B Connectivity distribution histogram and scale-free topology check when β = 6. C Gene dendrograms and clustering of module eigengenes using a dissimilarity metric (1-TOM) (the red line represents a cut height of 0.25). D Module-trait correlations between module eigengenes and sample traits were assessed. The correlation coefficient and P value are displayed in each cell. E Venn diagram showing where the DEGs and the red module overlap. F LASSO regression's partial likelihood deviance with changing log_(l) is shown in tenfold cross-validations. Using the minimum criterion (lambda.min) and 1 standard error of the minimum criterion (1-SE criteria), dotted vertical lines were created at the ideal values. G The tenfold cross-LASSO validation coefficient profiles for three hub genes are shown

Fig. 4

Confirmation of hub genes. A Boxplots in the training set were applied to validate the hub genes' expression levels. B ROC analysis in the training set was used to validate the diagnostic utility of the hub genes. C Boxplots in the test set were used to verify the hub genes' expression levels, and (D) ROC analysis in the test set was utilized to establish the hub genes' diagnostic value

Fig. 5

Analysis of the immunological landscape concerning disease severity. The distribution of immune cells in the IMV and NIMV groups is depicted in an (A) heatmap and (B) violin plot. C The connection between immune cell infiltration and hub genes