Based On Bioinformation Analysis,a Four-gene Diagnostic Model Of Sepsis Shock Was Constructed

Objective:Septic shock is a disease of organ failure caused by systemic infection with a predominance of shock.Its mortality rate is high,and the existing clinical criteria being used to determine it lack specificity.This study was conducted to construct molecular-level diagnostic markers using bioinformatics to accurately assess the individual status of patients.Methods:1.In this study,the chip dataset GSE95233 related to septic shock was downloaded from Gene Expression Omnibus（GEO）,and the data were corrected and screened with R packages（|log₂FC|>1.5 and adjust for differential genes（DEG）in P<0.05）.Subsequently,the gene was cut by WGCNA（weighted correlation network analysis）weighted gene co-expression network analysis method,and the gene module with the highest correlation with disease traits was taken as the core gene set,and some genes significantly associated with sepsis shock were found.The intersection of WGCNA and DEG was taken to obtain a significantly related and significantly differentially expressed gene in sepsis shock.2.The four machine learning algorithms of RF,GBM,XGBoost and SVM were used to compare the residuals,ROC curves,residual reverse distribution and characteristic gene importance of the genes screened in the 3,so as to select the best algorithm among the four algorithms,classify the intersecting genes by using the best algorithm among the four algorithms,and then use LASSO regression to screen the data dimensionality to construct diagnostic model genes.3.The expression difference of screening genes between sepsis shock and normal state was displayed by Boxplot plot,the composition of immune cells in GSE95233 was analyzed by CIBERSORT algorithm,and the correlation coefficient between four differential genes and immune cells was calculated by COR function,and the correlation between core differential genes and immune cells was observed.4.The nomogram of the diagnostic model of the gene selected by LASSO was constructed,and the CALIBRTION curve and the DCA decision curve were used to score the model to judge the practicability of the model.5.Download the single-cell sequencing data GSE167363 related to sepsis shock from GEO,and analyze these four genes from single-cell sequencing.After GSE167363 was quality controlled,the data were clustered by t SNE clustering algorithm,and then screened（|log₂FC|>1 and adjust the marker gene for P<0.05）for cell annotation.6.The expression of four genes in immune cells in GSE167363 was displayed by violin diagram,and then the expression trend of the four genes in the development trajectory of T cells and NK cells was observed through single-cell quasi-chronological analysis.Results:1.The GSE95233 data downloaded from GEO in this study included samples from 22 normal states and 102 cases with septic shock states.Differential analysis（|log₂FC|>1.5 and P<0.05）yielded 324 significant DEGs,including 133 downregulated genes and 191 upregulated genes associated with septic shock.In addition,according to WGCNA analysis,the genes of GSE95233 samples were divided into 9 modules,and the gene module ME Turquoise(cor=0.86,P=7e^-37)with the highest absolute correlation value was selected as the core gene set,and 2344 genes associated with sepsis shock were found,and the gene significance（GS）value and module member（MM）correlation was(cor=0.93,P<1e^-200).2.The 2344 genes screened by WGCNA and 324 DEGs were intersected to find 118significantly related and different genes.These 118 genes were run using four machine learning algorithms（RF,XGB,SVM,GLM）to find the optimal solution,and the RF algorithm was found to be the best algorithm.After the RF classifier was run,thirty genes with an importance ranking greater than 0 were obtained,and then four differential genes（DDX24,GUSBP2,P2RY10,SYNE2）were screened by LASSO regression.At the same time,it was found that the expression of these four genes in the transcriptome of sepsis shock showed a significant down-regulation trend（P<0.001）.3.The immunoinfiltration analysis of these four genes（DDX24,GUSBP2,P2RY10,SYNE2）in GSE95233 found that the contents of CD8⁺T cells,CD4⁺memory T cells,monocytes and M0 macrophages with normal and septic shock were quite different（P<0.001）,and the COR function was found to be positively correlated with CD8⁺T cells and CD4⁺memory T cells,and negatively correlated with M0 macrophages and plasma cells.4.These four genes（DDX24,GUSBP2,P2RY10,SYNE2）were constructed into a nomogram diagnostic model,and the model was scored by CALABRTION to find that the predicted value was close to the real value,and the DCA decision curve was carried out to find that the model curve was significantly higher than the two reference lines,indicating that this model has practicality.5.Download GSE167363 from GEO database,including a single-cell sequencing dataset of12 patients with sepsis,and integrate 6567 normal cells and 2457 sepsis cells after quality control,the cell plotting and clustering dendrogram shows that the resolution of 0.5 is good,and the t-SNE annotation after cell clustering shows that there are only 6 cell clusters respectively,B cells,granulocytes-macrophage progenitor cells,monocytes,NK cells,platelets and T cells.Conclusion:In this study,four differential genes（DDX24,GUSBP2,P2RY10,SYNE2）associated with sepsis between septic shock and normal state were identified and a diagnostic model for septic shock was constructed using WGCNA and machine learning on the basis of the GSE95233 transcriptomic dataset.The accuracy of this diagnostic model was demonstrated using the GSE131761 septic shock dataset,and analysis with the single-cell dataset GSE167363 suggested that the genes SYNE2 may play a key role in the diagnostic process of septic shock.