Research On GWAS Data Mining Of Alzheimer Disease Based On Protein-protein Interaction Network

Alzheimer’s disease(AD)is one of the most common neurodegenerative diseases.Its typical pathological features are amyloid deposition,neurofibrillary tangles,and senile plaque formation.With the increasing degree of population aging in the world,AD has not only affected the normal life of people,but also greatly increased the cost of social health care.It has become a serious public health problem.At present,the cause of Alzheimer’s disease is poorly understood,and there is no reliable indicator for early diagnosis.The current research on the early diagnosis of AD is mainly focused on the study of neuropsychological markers and biological markers,including cerebrospinal fluid(CSF)biomarkers,blood biomarkers,genetic markers,etc.In recent years,twin studies have confirmed that AD is highly heritable and is considered to be an important indicator of the earliest risk of disease.Genome-Wide Association Studies(GWAS)has been used as an important tool to study the pathogenesis of AD.Multiple risk genetic variants of AD have been discovered.Because of the "ignored" potential interactions between genes in the GWAS analysis,"lost" genetic loci occur.The protein-protein interaction(PPI)network is composed of protein interacting with each other to participate in biological signal transmission,gene expression regulation,energy and substance metabolism,cell cycle regulation and other life processes.Then,for GWAS data,the functional network modules are identified through PPI network analysis method and then performing functional annotations on them are closer to the theory of system biology.This strategy not only provides a novel perspective to understand the pathophysiology of AD,but also has important implications for establishing early diagnosis and clinical evaluation of other biomarkers.In view of this,this paper focuses on the strategy of mining AD GWAS data based on PPI networks.The GWAS data mining overlap function network module for CSF and imaging phenotypes is further studied.The main research contents of this article are:First,data preprocessing is performed.Samples are extracted and processed against the body fluid phenotype CSF of the ADNI database and the FreeSurfer data of the image phenotype MRI.Finally,843 samples of the CSF phenotype and 866 samples of the FreeSurfer are determined.The SNP(Single Nucleotide Polymorphisms)level analysis is performed on the identified samples and 56,3,980 SNPs are obtained to prepare the data for subsequent gene level experimental analysis.Secondly,for the t-tau GWAS data in the CSF,they are randomly grouped and identified as three sets of data,all of which are analyzed at the gene level.According to the characteristics of PPI networks,the traditional PageRank algorithm is improved.The degree of importance of protein nodes is introduced as a weight.A PPI network function module mining method based on weight-adjusted PageRank algorithm is designed to obtain a prioritized network.Then Jaccard’s similarity measure method and meta-analysis method are used to mine overlapping network function modules,and the overlapping network modules containing AD-related risk genes are found.In addition,for the CSF t-tau/Aβ1-42 GWAS data,the GWAS results are mapped to the PPI network to build the AD-disease network,and a consensus module mining algorithm CM-iPINBPA based on a restart random walk model is designed to integrate RWR,greedy algorithms,and consensus algorithms that mine consensus modules to increase algorithm stability.This method mines four consensus modules,validates these consensus modules(CMs)from a biological perspective,and discovers multiple genes with related biological functions.Finally,for the freesurfer multi-phenotype GWAS data in MRI,a MGAS model-based consensus network module mining algorithm(MGAS-CMs)is designed to mine the CMs related to image phenotype in PPI network.Using MGAS-CMs algorithm to mine five CMs related to the Freesurfer image phenotype.Multiple AD risk genes are found in the experiment,confirming that the statistical power combined with the multivariate genome-wide association analysis and the consensus network module mining idea is higher than the traditional GWAS method.In summary,the use of network analysis strategies to find the relationship between AD phenotype and genetic variation can not only consider the effect of each gene on the disease from the perspective of multiple phenotypic associations or interactions,but also help from the perspective of network modules and biological pathways and explain the molecular mechanisms of Alzheimer’s disease.