Essential blood molecular signature for progression of sepsis-induced acute lung injury: Integrated bioinformatic, single-cell RNA Seq and machine learning analysis.

Abstract

In this study, we aimed to identify an essential blood molecular signature for chacterizing the progression of sepsis-induced acute lung injury using integrated bioinformatic and machine learning analysis. The results showed that a total of 88 functionally related ALI-associated hub genes in sepsis were identified by MCODE analysis and they were enriched in infection and inflammtory responses, lung and cardiovascular disease pathways. These hub genes stratified ALI-sepsis and sepsis and further stratified two subtypes of sepsis-ALI with differential ALI scores, hub gene expression patterns, and levels of immune cells. A seven-gene signature including TNFRSF1A, NFKB1, FCGR2A, NFE2L2, ICAM1 and SOCS3 and PDCD1 was derived from the hub genes. These genes were significantly implicated in immune and metabolism pathways. They were expressed in six circulatory immune cells based on analysis of a single cell RNA sequencing dataset. Furthermore, the seven-gene signature was corrobarated using by integrating 12 machine learning algorithms. A premium three-gene signature NFE2L2, FCGR2A and PDCD1 for differentiating ALI-sepsis from sepsis were also derived from the seven-gene signature based on analysis of the seven core hub genes by the machine learning algorithms. Furthermore, the expressions of hub genes were verified in sepsis mice models. Therefore, our study provided an avenue to develop a molecular tool for identify and characterize progression of acute lung injury associated with sepsis.