Mining Alzheimer’s Disease Risk Factors By Integrating Multiple Omics Data Based On Mendelian Randomization

Alzheimer’s disease(AD)is a complex neurodegenerative disease.Due to the complexity of human cognition and brain functions,its pathogenesis is still unclear.Although genome-wide association study(GWAS)has successfully located more than 20 AD risk mutation sites,there are still long and complex pathways between the sites and the disease,including gene expression and protein.Downstream molecules have stronger effects on biological traits,and are worthy of in-depth research and analysis.The development of next-generation sequencing technology(NGS)and molecular quantitative detection technology has provided abundant data for biological and medical research.The use of statistical and computational methods to infer the association of biological macromolecules(genes or proteins)with phenotypes is becoming more and more promising,such as Mendelian randomization(MR)methods.In this study,we used the summary-data based MR(SMR)method to explore how genetic variation affects gene expression quantitative traits loci(eQTL),splicing QTL(sQTL),protein QTL(pQTL),and ultimately affects multiple complex phenotypic traits such as brain imaging of AD.In order to explore the genetic regulatory factors and even the regulatory mechanisms of specific brain regions from multiple omics levels,the SMR method was firstly used to analyze the GWAS results of the imaging phenotypes quantitative traits(iQTs)of 11 brain regions in the ENIGMA database and the eQTL/pQTL analysis results of the Dorsal Lateral Prefrontal Cortex(DLPFC)region from the ROSMAP database.It was found that the expression levels of 40 genes and the expression levels of 18 proteins were significantly associated with structural traits in these brain regions.Second,the TWAS and SMR methods were used in the Religious Orders Study and Memory and Aging Project(ROSMAP)database.The SMR method found that the expression and splicing levels of 70 genes have significant causal associations with the structural traits of the brain.And when using TWAS method to verify,there are 46 and 24 significant genes recurred respectively.Third,ROSMAP multi-omics data were combined with Jansen and Kunkle’s meta-GWAS analysis,among these 70 genes,we found 9 and 11 significant causally related genes and 23 genes were found using Bayesian Colocalization(coloc)method.Genes were analyzed by coloc colocalization(posterior probability ≥0.8).In order to verify the effectiveness of the SMR method,this article exploratively used a variety of MR frameworks to analyze the pleiotropy,heterogeneity,and sensitivity of the results.Finally,this research analyzed the protein-protein interaction network of the candidate gene set in the STRING database and found two closely connected clusters;and performed GO term enrichment analysis and KEGG pathway analysis,we then found a number of process Pathways and biological processes which were significantly associated with AD.In short,this research have integrated genetic,genomic,splicing,proteomic,and brain imaging omics information,focusing on a specific brain area,the Dorsal Lateral Prefrontal Cortex(DLPFC),and totally described the pathological changes of AD at a multi-omics molecular level.