Application of weighted co-expression network analysis and machine learning to identify the pathological mechanism of Alzheimer's disease

• Published in: Frontiers in aging neuroscience Vol. 14; p. 837770
• Main Authors: Chai, Keping, Zhang, Xiaolin, Chen, Shufang, Gu, Huaqian, Tang, Huitao, Cao, Panlong, Wang, Gangqiang, Ye, Weiping, Wan, Feng, Liang, Jiawei, Shen, Daojiang
• Format: Journal Article
• Language: English
• Published: Lausanne Frontiers Research Foundation 13-07-2022; Frontiers Media S.A
• Subjects: Aging Neuroscience; Algorithms; Alzheimer's disease; Biomarkers; Braak stages; Brain; Cognitive ability; Cortex (entorhinal); Cortex (frontal); Cortex (temporal); Gene expression; Gene set enrichment analysis; Inflammation; LAPTm5 protein; Learning algorithms; Machine learning; Microglia; neurodegeneration; Neurodegenerative diseases; Neurofibrillary tangles; random forest; ssGSEA; WGCNA
• ISSN: 1663-4365; 1663-4365
• DOI: 10.3389/fnagi.2022.837770

Aberrant deposits of neurofibrillary tangles (NFT), the main characteristic of Alzheimer's disease (AD), are highly related to cognitive impairment. However, the pathological mechanism of NFT formation is still unclear. This study explored differences in gene expression patterns in multiple brain regions [entorhinal, temporal, and frontal cortex (EC, TC, FC)] with distinct Braak stages (0- VI), and identified the hub genes
via
weighted gene co-expression network analysis (WGCNA) and machine learning. For WGCNA, consensus modules were detected and correlated with the single sample gene set enrichment analysis (ssGSEA) scores. Overlapping the differentially expressed genes (DEGs, Braak stages 0 vs. I-VI) with that in the interest module, metascape analysis, and Random Forest were conducted to explore the function of overlapping genes and obtain the most significant genes. We found that the three brain regions have high similarities in the gene expression pattern and that oxidative damage plays a vital role in NFT formation
via
machine learning. Through further filtering of genes from interested modules by Random Forest, we screened out key genes, such as LYN, LAPTM5, and IFI30. These key genes, including LYN, LAPTM5, and ARHGDIB, may play an important role in the development of AD through the inflammatory response pathway mediated by microglia.