Construction Of CeRNA Network And Prognostic Prediction Model Of Cervical Cancer Based On Bioinformatics

Purpose: This study used a bioinformatics approach to analyze the potential regulatory relationships between differentially expressed lncRNA,mRNA and mi RNA in cervical cancer and to construct a competitive endogenous RNA regulatory network in cervical cancer for the purpose of exploring the potential post-transcriptional regulatory mechanisms in cervical cancer.In this study,we attempted to find the RNA influencing the prognosis of cervical cancer patients and build a prognostic prediction model,so as to provide a new idea for the further study of molecular mechanism therapeutic targets and precision medicine for cervical cancer.Methods: RNA-seq and mi RNA-seq data and clinical information data of cervical cancer patients were downloaded from TCGA database and analyzed on R language software.The differentially expressed lncRNA,mRNA and mi RNA were screened by the fold change and false discovery rate,and the differentially expressed mRNA was verified in the three data sets of the GEO database.The GO gene function enrichment analysis and KEGG signaling pathway analysis of differentially expressed genes were performed by cluster Profiler package and SRTING database.The mi Rcode and DIANA Lnc Base databases were used together to predict the targeting interactions of mi RNAs and lncRNAs.star Base and mi RTar Base databases were used to predict the target genes of mi RNAs.The mi RNA-lncRNA and mi RNA-mRNA interactions were analyzed according to the ceRNA hypothesis theory,and significant lncRNA and mRNA pairs were identified.Finally,the lncRNA-mi RNA-mRNA ceRNA regulatory network of cervical cancer was constructed.Survival analysis of differentially expressed RNA was performed to identify prognostic related RNAs in the CERNA network and to construct a 5-year prognostic prediction model.The model was constructed using three algorithms respectively: random forest,support vector machine,and Cox risk proportional model based on LASSO algorithm feature selection.The prognostic performance of the three models in cervical cancer at 5 years was compared.A nomogram was drawn according to the Cox risk model.Results: This study screened out 491 differentially expressed mRNAs,191 differentially expressed mi RNAs and 241 differentially expressed lncRNAs.Differentially expressed genes were mainly enriched in biological functions such as chromosome segregation,nuclear segregation and nuclear chromosome segregation,as well as cell cycle,p53 and focal adhesion signaling pathways.Through the target relationship prediction results of 4RNA databases and the analysis of the GDCRNATools package,a competitive endogenous RNA network in cervical cancer comprised by 44 lncRNAs,27 mi RNAs and 42 mRNAs were finally constructed.The genes in the ceRNA network were mainly enriched in biological processes such as the G1/S transition of the mitotic cell cycle and the G1/S phase transition of the cell cycle,and are enriched in various cancer-related signal pathways.According to the protein interaction network,the hub gene and two potential protein complexes were identified.There were 15 prognostic-related RNAs in the ceRNA network,among which BCL2,AC008771.1 and SOX21-AS1 were independent prognostic factors.In the 5-year prognostic prediction model,the AUC of the random forest was 0.840,the Recall rate was 83.33%,and the F1-score was 0.8571.The AUC of the SVM-RBF was 0.868,the Recall rate was 88.89%,and the F1-score was 0.8889.The optimal lambda value of the LASSO regression model was 0.1378,and 16 features with non-zero coefficients were obtained.These16 characteristics were incorporated into the Cox risk proportional regression model to construct a risk function prediction model.The AUC of the test set was 0.833.The survival curves of the high and low risk groups were significantly different.Finally,a nomogram was drawn based on the Cox risk prediction model.The nomogram had a C-index of 0.775 in the test set,and had good predictive performance in the prediction of 1-year and 5-year survival rates.Conclusion: This study constructed a ceRNA regulatory network of lncRNA-mi RNA-mRNA for cervical cancer and identified three independent prognostic factors which provided reference for the research on the post-transcriptional regulation of RNA and survival prognosis of cervical cancer.The 16-RNA risk score prediction model has good performance in the 5-year survival outcome of cervical cancer patients.The nomogram based on Cox model could predict the 1-year and 5-year survival rates of cervical cancer patients,which provides a reference for individualized precise treatment of cervical cancer.