Identification and Experimental Validation of OS-Related Gene Sets Based on Integrated Analysis of Single-Cell and Bulk RNA Sequencing Data with Machine Learning in Patients with Sepsis

Abstract

Sepsis is a severe organ dysfunction syndrome caused by a dysregulated host response to infection, closely associated with poor prognosis. It disrupts the balance between oxidative and antioxidative systems, which may ultimately result in cellular dysfunction and death. However, the key regulatory genes involved in this process remain unclear and require further investigation. In this study, we analyzed a single-cell RNA sequencing dataset from the Single Cell Portal and an oxidative stress (OS) gene set from GeneCards. We employed multiple algorithms and correlation analysis to identify OS-related gene sets that were upregulated in sepsis. Subsequently, RNA expression datasets from the Gene Expression Omnibus were used to filter for overlapping genes that were upregulated in the sepsis group. Furthermore, we used three machine learning algorithms to identify the optimal characteristic genes and verified them with animal models. Analysis of both scRNA-seq and bulk RNA-seq datasets using various algorithms revealed a significant increase in OS activity scores following sepsis, with heterogeneity observed across different cell layers. TXN, NUDT1, MAPK14, and CYP1B1 were found to be closely associated with the elevated OS levels in sepsis. Furthermore, our animal experiments confirmed a significant increase in OS activity in septic mice, along with elevated expression of TXN, MAPK14, and CYP1B1. This study is the first to elucidate the heterogeneity of oxidative stress at the single-cell level in sepsis. The identification of TXN, MAPK14, and CYP1B1 as pivotal regulators of oxidative stress in sepsis highlights their potential as biomarkers and therapeutic targets.

Keywords: Machine learning; Oxidative stress; Sepsis; Single-cell RNA sequencing; TXN.