Bioinformatic Analysis And Construction Of Risk Prediction Models In The Transcriptome And Post-transcriptional Alternative Splicing Process In Glioblastoma Multiforme

Background Genomic alteration,especially the alteration of message RNA(m RNA)and long non-coding RNAs(lnc RNAs),in the level of transcriptome and post-transcription have been shown to play significant roles in the oncogenesis and progression of glioblastoma multiforme(GBM).The aim of our research is to identify potential m RNA or lnc RNA and alternative splicing events that are closely related to the prognosis of GBMMethods The profiling data of all RNA sequence from normal brain tissue or tumor tissue of patients with GBM were obtained from the Genotype-Tissue Expression(GTEx)and The Cancer Genome Atlas(TCGA).Differentially expressed gene identified were used to construct competing endogenous RNA(ce RNA)network and corresponding functional enrichment analysis was performed.To validate significant prognostic factors of GBM,univariate Cox regression followed by lasso and multivariate Cox analysis were used.Independent lnc RNA as potential prognostic factor were validated and a risk prediction model was constructed.Quantitative polymerase chain reaction(q PCR)was performed to verify the expression levels of potential lnc RNA biomarker in human GBM specimens.A gene set enrichment analysis(GSEA)was subsequently conducted to explore potential signaling pathways in which critical lnc RNAs may be involved.Moreover,a nomogram was applied based on our prediction model and significant clinical co-variate to visualize prognosis of GBM patients.TCGA Splice Seq data and the corresponding clinical data were downloaded from the TCGA data portal.Survival-related AS events were identified through Kaplan–Meier survival analysis and univariate Cox analysis.Experimentally validated or predicted splicing factors based on bioinformatic analysis that play important roles in the development and progression of malignant tumors were collected through literature or database retrieval.Then,splicing correlation network was constructed,and lasso regression analysis followed by multivariate Cox analysis were performed to validate independent AS as prognostic factor and a risk prediction model was constructed.Enrichment analysis was subsequently conducted to explore potential signaling pathways related to these AS events.Moreover,nomogram,a graphical plot which can be used for solving certain types of equations,was applied based on our risk prediction model and significant clinical co-variates to visualize the prognosis of GBM patients.Results In the transcriptome level,a total of 2023 differentially expressed genes(DEGs)including 56 lnc RNAs,1587 message RNAs(m RNAs)and 380 other RNAs were identified.Based on predictive databases,16 lnc RNAs,32 micro RNAs(mi RNAs)and 99 m RNAs were used to construct a ce RNA network.Moreover,we performed a novel risk prediction model with 5 potential prognostic related lnc RNAs,in which 4 of them were newly identified in GBM,to predict the prognosis of GBM patients.The nomogram was constructed,and the C-index is 0.774,indicating that the model has a good prediction effect.In the post-transcriptome level,a total of 45,610 AS events were included in our study,among which 416 survival-related AS events were identified.An AS correlation network including 54 AS events and 94 splicing factors,was constructed.Further functional enrichment analysis indicated that our AS correlation network was associated with cell growth,migration,apoptosis and tumor immune-related pathways.Moreover,area under curve(AUC)values of receiver operating curve(ROC)in the novel risk prediction model we constructed at one,two and three years were 0.953,0.973 and 0.816,respectively.The C-index of our nomogram is 0.733,which indicated that our risk prediction model performed superbly in prognosis prediction of GBM.Conclusions Differentially expressed genes in transcriptome and survivor-related splicing events in post-transcriptional process are important factors in biological function and prognosis prediction.Our findings in this study can deepen the understanding of the complicated mechanisms in transcriptional and post-transcriptional level of GBM and provide novel insights for further study.Moreover,it is envisioned that our risk prediction models and nomograms are used for preliminary clinical applications.Although the robustness of our models have been determined theoretically,more studies are needed for further optimization.