Research On The Method Of Identifying Biomarkers Based On Dynamic Network Entropy

Hepatocellular carcinoma(HCC)is one of the most common malignant tumors in the world.It is also one of the few cancers that has been observed to continue to rise in incidence over the past few years.Hundreds of millions of people worldwide suffer from this complex disease and its complications,and there is currently no effective treatment.Long-term interaction between lifestyle and genetic factors can cause HCC,but its pathogenesis is still not fully understood.A very important reason for the difficulty of early diagnosis and poor prognosis of HCC is its pathologic mechanism is still unclear.Most of the reported studies have only focused on the functions of individual differentially expressed genes in liver cancer tissues and the dysregulation of signaling pathways they participate in.However,the changes of complex regulatory networks during the occurrence and development of liver cancer are not limited to this.Molecular biomarkers are molecular indicators of specific biological conditions,such as normal and disease states.They are usually used to predict the diagnosis of disease,the risk assessment of treatment,and the treatment assessment of prognosis.Given that early detection of HCC can significantly improve the survival rate,it is particularly important to be able to find accurate biomarkers for the early diagnosis of HCC in the clinic.Gene expression profiling technology measures the transcription of thousands of genes in parallel.More and more liver cancer transcriptome data are published and available.The available high-throughput transcriptome data sets provide unprecedented opportunities for the discovery of liver cancer biomarkers.This thesis proposes a computational method based on a differential network entropy analysis framework,through a comprehensive analysis of human multi-stage HCC transcriptome,protein-protein interaction(PPI)and pathway data to identify potential HCC pathway biomarkers.First,we collect the pathways deposited in the databases and map them with the corresponding gene expressions in samples at different stages.They are linked by PPI as an interaction network.The pathway entropies are defined to evaluate their dysfunction-related activities and implications during the development and progression of HCC.We rank these pathways by their differential status of entropy dynamics in the time series of disease progression.Then,the pathway biomarkers are screened out by their classification ability of distinguishing the progression steps of HCC.The comparative studies with other alternative methods have proved the effectiveness and advantages of our proposed biomarker discovery strategy.The classification performance on independent data sets further validates the diagnostic applicability of these identified pathway biomarkers.Functional enrichment analysis of these pathway biomarkers also indicates the pathogenesis of HCC.