Abstract
<jats:p>Background: Parkinson&apos;s disease (PD) is a progressive neurodegenerative disorder characterized by the loss of dopaminergic neurons in the substantia nigra. While research into its pathogenesis has advanced, effective disease-modifying therapies remain elusive. Recently, through machine learning analysis of PD-related gene expression data, cytidine and deoxycytidine monophosphate (dCMP) deaminase domain containing 1 (CDADC1) has been identified as a potential therapeutic target.Methods: We developed a triple- artificial intelligence (AI) virtual screening platform integrating molecular docking, quantitative structure-activity relationship (QSAR) modeling, machine learning, and graph neural networks (GNN) for the identification of potential CDADC1-targeting compounds. The screening workflow was applied to a compound library containing 2,882 Food and Drug Administration (FDA)-approved drugs and 50,000 compounds from the ChEMBL database. Validation included absorption, distribution, metabolism, excretion, and toxicity (ADMET) prediction, molecular dynamics simulations, and comparative analysis with known PD therapeutics.Results: The triple AI screening strategy identified 15 high-confidence candidate compounds, which demonstrated strong binding affinity (binding free energy < −8.0 kcal/mol), high predicted bioactivity probability (> 0.8), and favorable drug-like properties. Among them, acetaminophen, amantadine, and minocycline exhibited the most outstanding overall scores. Molecular dynamics simulations confirmed stable binding between these compounds and the CDADC1 protein. Further cross‑validation analysis revealed that these compounds are potentially involved in PD‑related pathways, including the regulation of neuroinflammation and dopaminergic signaling.Conclusions: This study establishes a robust triple-AI virtual screening platform for PD drug discovery and identifies several drug repurposing candidates that warrant further experimental validation. The CDADC1-targeting strategy provides a novel therapeutic approach for the treatment of PD.</jats:p>