We aim to identify plasma proteomic biomarkers associated with neurodegenerative diseases using UK Biobank’s multi-omic resources, including proteomics, genomics, and transcriptomics.
Specifically, we will integrate these datasets using statistical and machine-learning approaches (e.g., random forest, LASSO regression) to identify molecular networks linking genetic risk to circulating protein alterations.
The study will:
1. Analyze associations between baseline plasma protein levels and the future incidence of neurodegenerative diseases such as Alzheimer’s disease, Parkinson’s disease, and amyotrophic lateral sclerosis (ALS).
2. Integrate genetic, transcriptomic, and proteomic layers to identify causal molecular pathways through correlation and mediation analyses.
3. Build predictive models for early disease risk using machine-learning approaches and interpret key predictors using explainable AI (SHAP-based feature attribution).
This work will clarify multi-omic mechanisms underlying neurodegeneration and provide candidates for early biomarkers and preventive targets.