Uncovering the potential role of oxidative stress in the development of periodontitis and establishing a stable diagnostic model via combining single-cell and machine learning analysis

Abstract

Background: The primary pathogenic cause of tooth loss in adults is periodontitis, although few reliable diagnostic methods are available in the early stages. One pathological factor that defines periodontitis pathology has previously been believed to be the equilibrium between inflammatory defense mechanisms and oxidative stress. Therefore, it is necessary to construct a model of oxidative stress-related periodontitis diagnostic markers through machine learning and bioinformatic analysis.

Methods: We used LASSO, SVM-RFE, and Random Forest techniques to screen for periodontitis-related oxidative stress variables and construct a diagnostic model by logistic regression, followed by a biological approach to build a Protein-Protein interaction network (PPI) based on modelled genes while using modelled genes. Unsupervised clustering analysis was performed to screen for oxidative stress subtypes of periodontitis. we used WGCNA to explore the pathways correlated with oxidative stress in periodontitis patients. Networks. Finally, we used single-cell data to screen the cellular subpopulations with the highest correlation by scoring oxidative stress genes and performed a proposed temporal analysis of the subpopulations.

Results: We discovered 3 periodontitis-associated genes (CASP3, IL-1β, and TXN). A characteristic line graph based on these genes can be helpful for patients. The primary hub gene screened by the PPI was constructed, where immune-related and cellular metabolism-related pathways were significantly enriched. Consistent clustering analysis found two oxidative stress categories, with the C2 subtype showing higher immune cell infiltration and immune function ratings. Therefore, we hypothesized that the high expression of oxidative stress genes was correlated with the formation of the immune environment in patients with periodontitis. Using the WGCNA approach, we examined the co-expressed gene modules related to the various subtypes of oxidative stress. Finally, we selected monocytes for mimetic time series analysis and analyzed the expression changes of oxidative stress genes with the mimetic time series axis, in which the expression of JUN, TXN, and IL-1β differed with the change of cell status.

Conclusion: This study identifies a diagnostic model of 3-OSRGs from which patients can benefit and explores the importance of oxidative stress genes in building an immune environment in patients with periodontitis.

Keywords: WGCNA; diagnostic signature; inflammation; machine learning; oxidative stress; periodontitis; single-cell RNA-seq.