Abstract
PURPOSE: Sepsis is the 10th leading cause of death globally and the most common cause of death in patients with infections. Ubiquitination plays a key role in regulating immune responses during sepsis. This study combined bioinformatics and Mendelian randomization (MR) analyses to identify ubiquitin-related genes (UbRGs) with unique roles in sepsis. METHODS: Relevant genes were obtained from the GSE28750 dataset and GSE95233, weighted gene co-expression network analyses were performed to identify gene modules, and differentially expressed UBRGs (DE-UBRGs) were generated by differentially expressed genes (DEGs) crossover with key modular genes and UBRGs in sepsis and normal samples. Causal relationships between sepsis and UbRGs were analysed using MR, performance diagnostics were performed using subject work characteristics (ROC) curves, and an artificial neural network (ANN) model was developed. On this basis, immune infiltration was performed and the expression of key genes was verified in animal models. RESULTS: 3022 DEGs were found between sepsis and normal. A total of 2620 genes were obtained as key modular genes. Crossing DEGs, key modular genes and UBRGs yielded 93 DE-UBRGs. MR results showed WDR26 as a risk factor for sepsis (OR>1) and UBE2D1 as a protective factor for sepsis (OR<1), which was reinforced by scatterplot and forest plot. ROC curves showed that WDR26 and UBE2D1 could accurately differentiate between sepsis and normal samples. Confusion matrix and ROC curve results indicate that the artificial neural network model has strong diagnostic ability. The results of immune infiltration showed that.WDR26 was negatively correlated with plasma cells, while UBE2D1 was positively correlated with CD4 naïve T cells. Significant differences between sepsis and normal were obtained between UBE2D1 and WDR26 in the animal model. CONCLUSION: There appeared to be a causal relationship between sepsis, WDR26 and UBE2D1. The insights were of value for effective clinical diagnosis and treatment in sepsis.