Mining And Analyzing Potential Biomarkers Of Alzheimer’s Disease Based On Bioinformatics

Alzheimer’s disease(AD)is a neurodegenerative disease with clinical manifestations of impaired memory and other cognitive functions,which seriously affects the daily life of patients.There are only a few drugs approved to treat AD,but these drugs only relieve symptoms,don’t alter the course of the disease.Evidence suggests that the pathogenesis of AD is progressive,with AD-related pathological changes occurring before clinical symptoms,and drug therapy in the preclinical stage may be more effective.Therefore,the discovery of new AD-related biomarkers is particularly urgent for early diagnosis and treatment.In the course of AD,patients can go from normal to mild cognitive impairment and eventually to AD.In this study,we downloaded the gene chip data GSE140829 and corresponding clinical information from NCBI database.The data samples were divided into Control(CTL),Mild cognitive impairment(MCI)and AD.Firstly,differentially expressed genes in different disease states(MCI vs.CTL and AD vs.CTL)were obtained through gene expression differential analysis.Secondly,based on the differentially expressed genes,total module was analyzed parallelly using the weighted total gene network and consistent network.,,We found several modules with high coincidence degree and associated with AD from the function of information enrichment.,Then we calculated the correlation between modules and genes.12 and 10 AD-related genes were screened on the basis of sorting,and9 overlapping genes were further analyzed.EEF1B2 and ZNHIT3 have been experimentally confirmed to be associated with AD and a variety of neurological diseases,while the functions of TMEM126 B,LSM1,ZNF22,MRPL22,COMMD3,RWDD1 and MRPL32 have not been clearly reported.Then,based on module genes and 7 core genes,the normal state and disease state(MCI,AD)were classified and predicted by Support Vector Machine(SVM)and Neural network(NN)respectively.The results showed that the accuracy of genes and core genes in the module was up to 95% and 70%,which can effectively distinguish samples at different stages.The SVM model was also constructed from GSE63060 chip data,and the classification prediction accuracy was up to 96% and 75%.The prediction results of various data showed that the core genes had universality in AD differentiation.Finally,conservative analysis was used to calculate the stability of the functional module where the core genes were located,and it was confirmed that these 7 genes were dominant in their respective functional modules with high connectivity and strong correlation,proving their importance in AD.