Revealing novel pyroptosis-related therapeutic targets for sepsis based on machine learning

BMC Medical Genomics

Research

Xianyao Wan¹  Ying Chen²  Xingkai Wang³  Junwei Zong⁴  Jiaxin Wang⁵ 

[1] Department of Critical Care Medicine, The First Affiliated Hospital of Dalian Medical University, Dalian, Liaoning, China;Department of Critical Care Medicine, The First Affiliated Hospital of Dalian Medical University, Dalian, Liaoning, China;Department of Respiratory Medicine, The Second Affiliated Hospital of Dalian Medical University, Dalian, Liaoning, China;Department of Orthopedics, The First Affiliated Hospital of Dalian Medical University, Dalian, Liaoning, China;Department of Orthopedics, The First Affiliated Hospital of Dalian Medical University, Dalian, Liaoning, China;Institute (College) of Integrative Medicine, Dalian Medical University, Dalian, Liaoning, China;Institute (College) of Integrative Medicine, Dalian Medical University, Dalian, Liaoning, China;

关键词: Sepsis;    WGCNA;    Pyroptosis;    Machine learning;    Prediction model;   

DOI  :  10.1186/s12920-023-01453-7

 received in 2022-10-08, accepted in 2023-02-06,  发布年份 2023

来源: Springer

【 摘 要 】

BackgroundSepsis is one of the most lethal diseases worldwide. Pyroptosis is a unique form of cell death, and the mechanism of interaction with sepsis is not yet clear. The aim of this study was to uncover pyroptosis genes associated with sepsis and to provide early therapeutic targets for the treatment of sepsis.MethodsBased on the GSE134347 dataset, sepsis-related genes were mined by differential expression analysis and weighted gene coexpression network analysis (WGCNA). Subsequently, the sepsis-related genes were analysed for enrichment, and a protein‒protein interaction (PPI) network was constructed. We performed unsupervised consensus clustering of sepsis patients based on 33 pyroptosis-related genes (PRGs) provided by prior reviews. We finally obtained the PRGs mostly associated with sepsis by machine learning prediction models combined with prior reviews. The GSE32707 dataset served as an external validation dataset to validate the model and PRGs via receiver operating characteristic (ROC) curves. The NetworkAnalyst online tool was utilized to create a ceRNA network of lncRNAs and miRNAs around PRGs mostly associated with sepsis.ResultsA total of 170 genes associated with sepsis and 13 hub genes were acquired by WGCNA and PPI network analysis. The results of the enrichment analysis implied that these genes were mainly involved in the regulation of the inflammatory response and the positive regulation of bacterial and fungal defence responses. The prolactin signalling pathway and IL-17 signalling pathway were the primary enrichment pathways. Thirty-three PRGs can effectively classify septic patients into two subtypes, implying that there is a reciprocal relationship between sepsis and pyroptosis. Eventually, NLRC4 was considered the PRG most strongly associated with sepsis. The validation results of the prediction model and NLRC4 based on ROC curves were 0.74 and 0.67, respectively, both of which showed better predictive values. Meanwhile, the ceRNA network consisting of 6 lncRNAs and 2 miRNAs was constructed around NLRC4.ConclusionNLRC4, as the PRG mostly associated with sepsis, could be considered a potential target for treatment. The 6 lncRNAs and 2 miRNAs centred on NLRC4 could serve as a further research direction to uncover the deeper pathogenesis of sepsis.

【 授权许可】

CC BY   
© The Author(s) 2023

【 图 表 】

Fig. 2

Fig. 5

Fig. 4

Fig. 3

Fig. 1

Fig. 2

Fig. 2

42004_2023_830_Article_IEq3.gif

12302_2023_718_Article_IEq50.gif

Fig. 6

Fig. 4

Fig. 3

Fig. 2