Biomarkers Network Modeling Of Several Cancers

Cancer is the most common type of malignant tumor,which does great harm to human health.The incidence of most cancers is insidious,and the rate of early clinical diagnosis is low,so there is an urgent need for early cancer screening techniques with high accuracy.Combining gene technology and multi-omics biological big data to screen cancer biomarkers can not only systematically study the biomolecular mechanism,but also reveal the molecular regulation mechanism from multiple perspectives.However,in the face of massive high-throughput data,its processing and application become a vital step.Network modeling can better reveal the interaction between molecules,and show the molecular regulation system through the network diagram.Based on the idea of network modeling,this paper mainly studies the biomarkers screening of disease gene-gene network,disease evolution mechanism based on dynamic network and the extraction of dynamic biomarkers,the relationship between cancer key mutation genes and prognosis analysis.The main contents of this article are as follows:In the second chapter,in order to distinguish and predict cancer,a new method of screening biomarkers based on gene-gene interaction network and feature extraction algorithm was proposed,which combined graph theory and machine learning optimization,and its purpose was to extract biomarkers with high accuracy and as few biomarkers as possible.First,the data of cancer-related gene expression and mi RNA expression were used to obtained differential genes by differential expression analysis(DEA),and then we constructed a gene-gene multi-interaction network.Furthermore,combination screening algorithm was applied to choose the optimal biomarker gene set.The method was applied to LUAD dataset,and six key biomarkers were obtained,including CAV1,CCNA2,EZH2,RBP4,ITIH1 and EGFR,especially CCNA2 as a potential biomarker.Finally,four independent datasets was applied to test the robustness.High precisions(ranging from 94.4% to 100%)showed that our method can be effectively distinguished and predicted cancer in the clinical stage.In the third chapter,considering that the development of disease is a dynamic process,therefore,a dynamic network model of disease development was constructed,which was named as the dynamic pathway evolution analysis(DPE-Analysis).The model can not only describe the evolution of a disease and determine its critical period,but also identify the driving genes through network traceability.Specifically,the directed regulation network was constructed by combining time-series datasets and pathway regulation data;then we proposed MCD-IHC algorithm and CCI criterion;finally,the disease critical period was determined by the improved dynamic network biomarker(DNB)algorithm.The model was applied to the time-series datasets of liver cancer,effectively dividing it into three stages,and the driving biomarkers were JUNB,MYH7,MAP2K6,HSPA1 B and COL15A1.This method can help researchers describe the disease progress and provide research ideas for screening driving biomarkers.In the fourth chapter,considering that the biomarkers screened in the first two chapters were all biological macromolecules such as genes and m RNAs with dysregulated expression,in order to explore the impact of biomarkers on the prognosis of patients,only disease mutation genes were studied in this chapter.We proposed a new method combining social network analysis with Floyd algorithm to construct mutant gene-gene interaction network,and we defined C-score criterion to obtain important mutant genes,further use them as seed nodes to reconstruct protein-protein interaction network(PPIN),then verify the validity of key mutant genes through KEGG and GO analysis.Survival analysis was performed using Kaplan-Meire method and Tarone-Ware test.This method was applied to breast cancer and found that the combination of the TP53-GATA3-PIKCA gene-set has a significant effect on the survival of breast cancer patients.Then the test of LUAD and liver cancer data validated the significant impact of biomarkers on patient prognosis.Therefore,it can be considered for further clinical validation.