Machine Learning Identifies TRAPPC13/COPS5 as Biomarkers and Vesicle Transport Subtypes in Parkinson's Disease

Area of Science:

• Neuroscience
• Genomics
• Biomarker Discovery

Background:

• Parkinson's disease (PD) involves neuron loss and protein trafficking issues.
• Vesicle-mediated transport dysregulation is implicated in PD pathogenesis.
• Diagnostic and immune roles of vesicle transport in PD remain unclear.

Purpose of the Study:

• Identify robust diagnostic biomarkers for Parkinson's disease.
• Investigate immune associations and molecular subtypes in PD.
• Evaluate the diagnostic and therapeutic potential of identified biomarkers.

Main Methods:

• Analyzed transcriptomic data from multiple GEO datasets.
• Employed machine learning (LASSO, Random Forest, XGBoost) to identify diagnostic genes.
• Validated the diagnostic model and assessed immune infiltration using NMF.

Main Results:

• TRAPPC13 and COPS5 identified as robust diagnostic biomarkers for PD with high accuracy.
• Diagnostic model showed excellent predictive accuracy, calibration, and clinical applicability.
• Identified two distinct PD molecular subtypes with unique pathway and immune signatures.

Conclusions:

• TRAPPC13 and COPS5 are novel, highly accurate diagnostic biomarkers for Parkinson's disease.
• The identified PD subtypes offer new insights into disease pathogenesis.
• These findings may facilitate personalized therapeutic approaches for PD patients.