Abstract
Parkinson’s disease (PD) is a neurodegenerative disorder linked to alpha-synuclein pathology and dopaminergic neuron loss. Exosomes play dual roles in PD progression, facilitating pathological protein spread or exerting neuroprotective effects. This study aimed to identify exosome-related gene biomarkers for PD diagnosis and therapy. Using multi-dataset analysis (Series: GSE42966, Platforms: GPL4133, Series: GSE99039, Platforms: GPL570, Series: GSE156926, Platforms: GPL19920) from the Gene Expression Omnibus (GEO), we integrated principal component analysis (PCA), differential gene screening (|logFC|>0.5, p < 0.05), and machine learning to identify PD-associated exosome-related genes. Functional enrichment revealed associations with oxidative phosphorylation, Huntington’s disease pathways, and immune responses. A predictive model comprising five genes—GNAS, TUBB2A, RPL22, RPL5, and WNT5A—was established and validated using ROC curves (Receiver Operating Characteristic Curve) and nomograms. Immune profiling linked these genes to B cells, MDSCs, and CD4+/CD8 + T cells. Drug-gene network analysis highlighted interactions with compounds like NSC94017 and phosphine, while molecular docking identified key binding residues (e.g., GLN294, ASP295). These genes may serve as early diagnostic biomarkers and therapeutic targets. Despite promising results, further validation in larger cohorts and mechanistic studies are needed to confirm their roles in PD pathogenesis and treatment. This study provides a foundation for developing precision gene therapies and non-invasive diagnostic strategies for PD. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s12883-025-04477-x.