Construction Of Tumor Pre-deterioration Critical Stage Recognition Model Based On Single Sample Node Entropy And Its Application In Identification Of Dynamic Network Biomarkers

Most progressive malignancies have low five-year survival rates.Early identification?diagnosis and intervention will greatly improve the chances of survival of patients.During the progression of cancer,there is always a critical stage before deterioration,which is difficult to be detected but separates the steady and the progressive stage of the cancer.This stage provides a time window for clinical treatment and important clues for drug research.However,how to identify this important stage from the chaotic clinical data of cancer cohort remains a challenge.In this study,we successfully constructed an algorithmic based on single-sample node entropy and performed analysis of publicly available cancer RNA sequencing data in TCGA.Taken together,the results of this study are described below:(1)In this study,using intergenic interactions network as a signaling network,a novel single-sample model based on the concept of entropy in informatics was built on the basis of the existing computational framework for dynamic network biomarkers.This model was expected to be able to identify the critical stage before cancer progression based on time-series data from an individual cancer sample and to uncover the biological mechanisms that drive the disease past that critical period.(2)This study validated the usability of the model by using it in a simulation data set.At the same time,we established the association between critical stage of cancer development and patients' overall survival in four cancer cohort data.(3)In this study,we compared the model with the results of the differential expression analysis and found that most of the core genes discovered using node entropy occupied upstream in important cancer-associated signaling pathways,while differential expressed genes in the same pathway were in downstream.(4)Our study also found that changes in cancer-type-specific core gene networks derived the progression of different cancers.These core genes have the potential to be drug targets for different cancers.Furthermore,we found that non-specific pathways across cancer types were strongly associated with DNA repairing.In these pathways,the switch of expression patterns of high nodal entropy genes and their interacting gene was a major force for cancer progression.(5)Finally,this study also used nodal entropy to identify some novel prognosis biomarkers such as PER2,TNFSF4,MMP13,ENO4,etc.,and explored potential mechanisms for some genes' affection on prognosis.In conclusion,we have successfully built a dynamic single-sample node entropy model and used it to find the tipping point for cancer development and possible hidden mechanisms promote the progression.These findings may provide critical timing reference and new target genes for clinical medical intervention strategies and drug development.