Construction of gene network database, and identification of key genes for diagnosis, prognosis, and treatment in sepsis

Abstract

Sepsis and sepsis-related diseases cause a high rate of mortality worldwide. The molecular and cellular mechanisms of sepsis are still unclear. Six sepsis-related blood transcriptome datasets were collected and analyzed by weighted gene co-expression network analysis (WGCNA). Functional annotation was performed in gProfiler tool. DSigDB was used for drug signatures enrichment analysis. The proportion of immune cells was estimated by the CIBERSORT tool. The relationships between modules, immune cells, and survival were identified by correlation analysis and survival analysis. A total of 37 stable co-expressed gene modules were identified. These modules were associated with the critical biology process in sepsis. Four modules can independently separate patients with long and short survival. Three modules can recurrently separate sepsis and normal patients with high accuracy. Some modules can separate bacterial pneumonia, influenza pneumonia, mixed bacterial and influenza A pneumonia, and non-infective systemic inflammatory response syndrome (SIRS). Drug signature analysis identified drugs associated with sepsis, such as testosterone, phytoestrogens, ibuprofen, urea, dichlorvos, potassium persulfate, and vitamin B12. Finally a gene co-expression network database was constructed (https://liqs.shinyapps.io/sepsis/). In conclusion, the recurrent modules in sepsis may facilitate disease diagnosis, prognosis, and treatment.