Study on the Mechanism of Multigene Synergism in Alzheimer’s Disease Based on UMAP Dimensionality

Abstract

Alzheimer’s disease (AD) is a highly genetically complex neurodegenerative disease with a heritability of up to 76%. Current genome-wide association studies (GWAS) mainly focus on the significant effects of single genes (such as APOE ε4), but there is still a gap in the study of synergy between genes. This study proposes an integrative analysis framework based on the unified manifold approximation and projection (UMAP) method to reveal the hidden synergistic gene network in AD. We used peripheral blood whole genome sequencing data (Illumina Omni 2.5M BeadChip) from the ADNI2 (2010-2012) and ADNIGO cohorts and screened high-confidence samples through strict quality control (RIN ≥ 7). The results showed that two synergistic gene modules (Cluster 0: TNFRSF10D/OCLN/NIPAL2/BTNL3; Cluster 4: OLR1/MMP8/UTY/SCN3A) were identified by machine learning and gene cluster analysis, which showed high functional convergence in pathways such as neuroinflammation, vascular dysfunction, and neuronal apoptosis. Among them, the spatiotemporal heterogeneity of NIPAL2 expression in the temporal lobe region (maximum expression change: 1.63, temporal variance: 0.157) and the role of OCLN in the regulation of blood-brain barrier integrity (maximum expression change: 1.99) were particularly critical.