Elucidating sleep disorders: a comprehensive bioinformatics analysis of functional gene sets and hub genes

Abstract

Background: Sleep disorders (SD) are known to have a profound impact on human health and quality of life although their exact pathogenic mechanisms remain poorly understood.

Methods: The study first accessed SD datasets from the GEO and identified DEGs. These DEGs were then subjected to gene set enrichment analysis. Several advanced techniques, including the RF, SVM-RFE, PPI networks, and LASSO methodologies, were utilized to identify hub genes closely associated with SD. Additionally, the ssGSEA approach was employed to analyze immune cell infiltration and functional gene set scores in SD. DEGs were also scrutinized in relation to miRNA, and the DGIdb database was used to explore potential pharmacological treatments for SD. Furthermore, in an SD murine model, the expression levels of these hub genes were confirmed through RT-qPCR and Western Blot analyses.

Results: The findings of the study indicate that DEGs are significantly enriched in functions and pathways related to immune cell activity, stress response, and neural system regulation. The analysis of immunoinfiltration demonstrated a marked elevation in the levels of Activated CD4+ T cells and CD8+ T cells in the SD cohort, accompanied by a notable rise in Central memory CD4 T cells, Central memory CD8 T cells, and Natural killer T cells. Using machine learning algorithms, the study also identified hub genes closely associated with SD, including IPO9, RAP2A, DDX17, MBNL2, PIK3AP1, and ZNF385A. Based on these genes, an SD diagnostic model was constructed and its efficacy validated across multiple datasets. In the SD murine model, the mRNA and protein expressions of these 6 hub genes were found to be consistent with the results of the bioinformatics analysis.

Conclusion: In conclusion, this study identified 6 genes closely linked to SD, which may play pivotal roles in neural system development, the immune microenvironment, and inflammatory responses. Additionally, the key gene-based SD diagnostic model constructed in this study, validated on multiple datasets showed a high degree of reliability and accuracy, predicting its wide potential for clinical applications. However, limited by the range of data sources and sample size, this may affect the generalizability of the results.

Keywords: diagnostic model; drugs; functional gene sets; hub genes; sleep disorders.

Figure 1

DEGs and enrichment analysis of SD. (A, B) Volcano plot of DEGs between Sleep disorders and Normal groups in GSE208668 and GSE240856. (C, D) Heatmap of top 20 DEGs in GSE208668 (C) and GSE240856 (D). (E, F) The top 10 gene sets that areactivated or inhibited in the C2 (E) and C5 (F) gene sets of MSigDB.

Figure 2

Immune Cell Infiltration (A, B) Violin plot comparing the results of two immune infiltration algorithms. ssGSEA (A) and ImmuCellAI (B). (C, D) Heatmap of the proportions of two immune infiltration algorithms in the Sleep disorders and Normal groups. ImmuCellAI (C) and ssGSEA (D). * p < 0.05, ** p < 0.01, *** p < 0.001.

Figure 3

Gene set scoring. (A) Correlation graph for 28 types of immune cells. (B) Heatmap of scores for 50 gene sets. (C) Box plot comparing scores of 12 signaling pathways between Sleep disorders and Normal groups. (D) Box plot comparing scores of 22 physiological functions between Sleep disorders and Normal groups. * p < 0.05, ** p < 0.01, *** p < 0.001.

Figure 4

(A) Venn diagram of differential genes between datasets GSE20851 and GSE208668. (B) Identification of genes in the PPI network common to differential genes. (C).Top 20 genes selected by the CytoHubba plugin.

Figure 5

Gene selection through machine learning. (A) The correlation plot between the number of Random Forest trees and model error. (B) Top 20 genes selected by the RF method. (C) Top 25 genes identified by SVM, ranked by the percentage decrease in mean impurity. (D) Results obtained from the predictive model of the Root Mean Square Error through cross-validation. (E, F) Cvfit and lambda curves demonstrating the use of the LASSO regression, performed with the minimum criteria.

Figure 6

Construction of a research and diagnostic model based on hub genes. (A) Expression of 6 hub genes in dataset GSE208668. (B) Correlation between 6 key genes and crucial signaling pathways. (C) A nomogram model, incorporating 6 hub genes, was constructed to predict risk. (D) The calibration curve of the nomogram to test the predictive performance of the model. (E) ROC curves analysis of GSE208668 for the diagnostic model. (F–H) display ROC curve analyses of the diagnostic model applied to GSE240851, GSE98582, and GSE56931 datasets. * p < 0.05, ** p < 0.01, *** p < 0.001.

Figure 7

Potential therapeutic drug search based on miRNA. (A) Volcano plot of miRNA differential analysis results from the GSE165041 dataset. (B) Venn diagram of DEmiRNAs and miRNAs obtained from miRNANet. (C) The Sankey plot shows the relationships between 5 miRNAs and their target genes. (D) Interaction scores between genes and drugs were obtained from the DGIdb database.

Figure 8

Expression of mRNA and proteins in mice with SD. (A) Relative mRNA expression of the 6 hub genes. (B) Western blot results for 3 relative proteins. (C) Relative protein expression. * p < 0.05, ** p < 0.01, *** p < 0.001.