Exploration Of The Molecular Mechanism And Prognosis Risk Model Of Ewing’s Sarcoma Based On Bioinformatics Analysis

Objective: 1.To construct a scale-free network by weighted gene co-expression network analysis(WGCNA)based on the differentially expressed gene(DEGs)of Ewing’s sarcoma(ES)to find hub genes related to clinical prognosis of ES patients,and the biological functions.2.Based on COX analysis,a prognosis risk model for ES patients was constructed.3.Multiple expression profile chips of ES were combined with bioinformatics analysis to explore the hub gene of ES and its related biological functions.Methods:1.ES gene expression profiles and its clinically relevant information were downloaded from Gene Expression Omnibus(GEO)database data.ES expression profile chips were followed as the qualifications:(a)human ES patients mRNA expression profile chip raw data,not animal models,(b)containing ES tissue and normal control tissue or the expression profile chips of ES patients with detailed clinical data.2.Robust Multiarray Averaging(RMA)was performed to the preprocessing of raw data.The oligo package was used for GSE68776 and GSE6315,and the affy package was used for GSE17679 and GSE45544.3.The limma package was used to screen DEGs of GSE17679 following as the thresholds : P value <0.01,log2FC>1 or log2FC<-1,wherein up-regulated gene log2FC>1 and down-regulated gene log2FC<-1.4.The DEGs of GSE17679 were used to construct a co-expression network by using the WGCNA package and then to identify hub genes and its associated with clinical prognosis modules in ES patients and to explore possible biological functions of hub genes.5.The expression profile data of hub genes screened from WGCNA were performed to univariate COX analysis by using survival package and then were used to construct Robust likelihood-based survival model by using rbsurv package to choose genes,which were subjected to multivariate COX analysis and to construct survivalrelated linear risk assessment models.GSE63157 was used to verify the stability and the value of the model for ES.5.The sva package was used for removing batch effection of of GSE17679,GSE63157 and GSE68776.6.The DEGs of multiple ES expression profiles were screened with the limma package(DEGs screening following as the threshold: adj.P value < 0.01,log2FC>1.5 or log2FC<-1.5,wherein up-regulated log2FC>1.5,down-regulated gene log2FC<-1.5)and explore its biological function and screen out hub genes.RESULTS: 1 Four ES-related expression profile such as GSE68776,GSE45544,GSE17679 and GSE63157 were downloaded from the GEO database.The GSE17679 experimental platform is GPL570,which includes 88 ES patient samples with clinical information,11 ES cell line samples and 18 normal control tissue samples.The GSE45544 experimental platform is GPL6244,including 14 ES patient samples and 8 ES cell line samples.And 22 normal control tissue samples;GSE68776 experimental platform is GPL5175,including 32 ES patient samples 91 ES cell line samples and 33 normal control tissue samples;GSE63157 experimental platform is GPL5175,including 88 clinical information ES patients samples.2.Under the threshold,4131 DEGs(1275 down-regulated,2856 up-regulated)were chosen from GSE17679 and used to WGCNA.The WGCNA screened four survival time-related modules,and 92 hub genes,GO enrichment analysis and KEGG signaling pathway analysis shown these hub genes are primarily involved in cell signal transport.3.COX analysis obtained a linear model: Risk score=CSPG5*(-4.2210)+DAPK1*(-3.4872)+DUSP13*(5.4790)+NPY1R*(1.6836)+OPTN*(-3.0195);the five-year survival time of the low-risk group(86.4%,95% CI=76.%8-97.1%)was significantly higher than the high risk(6.53%,95%CI=1.80%-23.7%)(P = 9.675e-12).The gene expression data and clinical data of ES patients in GSE63157 verified the validity of the model and proved that the model is a prognostic factor independent of other factors.4.Multi-microarray analysis screened out 574 DEGs(319 down-regulated genes and 255 up-regulated genes),and GO enrichment analysis and KEGG signaling pathways showed that these DEGs were mainly involved in muscle contraction,myofibril,actin binding and Cardiac muscle contraction.We screened 10 hub genes,which were mainly related to cell division.Conclusion: 1.Multi-microarray analysis,WGCNA,COX analysis and other robust and reliable bioinformatics analysis methods help ES understand the possible molecular mechanism and provide theoretical guidance for basic research of ES.2.The COX risk model based on 5 mRNAs can better predict the overall survival prognosis of ES patients.The 5 mRNAs are considered as independent prognostic markers for ES patients,and provided indicators and potential therapeutic targets and used to screen patients with high-risk,and guide for clinician treatment.