Identification of Methylated Gene Biomarkers in Patients with Alzheimer's Disease Based on Machine Learning. (27th March 2020)

• Record Type: Journal Article
• Authors: Ren, Jianting; Zhang, Bo; Wei, Dongfeng; Zhang, Zhanjun
• Other Names: Baloyannis Stavros Academic Editor.
• Is Part Of: BioMed research international. Volume 2020(2020)
• Publication Date: 2020-03-27
• Subjects: Medicine -- Periodicals; Biology -- Periodicals; Biotechnology -- Periodicals; Life sciences -- Periodicals; 610.5
• DOI: 10.1155/2020/8348147
• Languages: English
• ISSNs: 2314-6133

Abstract

Abstract : Background . Alzheimer's disease (AD) is a neurodegenerative disorder and characterized by the cognitive impairments. It is essential to identify potential gene biomarkers for AD pathology. Methods . DNA methylation expression data of patients with AD were downloaded from the Gene Expression Omnibus (GEO) database. Differentially methylated sites were identified. The functional annotation analysis of corresponding genes in the differentially methylated sites was performed. The optimal diagnostic gene biomarkers for AD were identified by using random forest feature selection procedure. In addition, receiver operating characteristic (ROC) diagnostic analysis of differentially methylated genes was performed. Results . A total of 10 differentially methylated sites including 5 hypermethylated sites and 5 hypomethylated sites were identified in AD. There were a total of 8 genes including thioredoxin interacting protein (TXNIP), noggin (NOG), regulator of microtubule dynamics 2 (FAM82A1), myoneurin (MYNN), ankyrin repeat domain 34B (ANKRD34B), STAM-binding protein like 1, ALMalpha (STAMBPL1), cyclin-dependent kinase inhibitor 1C (CDKN1C), and coronin 2B (CORO2B) that correspond to 10 differentially methylated sites. The cell cycle (FDR = 0.0284087 ) and TGF-beta signaling pathway (FDR = 0.0380372 ) were the only two significantly enriched pathways of these genes. MYNN was selected as optimal diagnostic biomarker with great diagnostic value. The random forests model could effectively predict AD. Conclusion . Our study suggested that MYNN could be served as optimal diagnostic biomarker of AD. Cell cycle and TGF-beta signaling pathway may be associated with AD.