Alzheimer's Disease Protein Relevance Analysis Using Human and Mouse Model Proteomics Data
Cathy Shi1, W Kirby Gottschalk2, Carol A Colton2
1Department of Statistical Science, Duke University, Durham, NC 27708, USA.
Frontiers in Systems Biology
|August 31, 2023
Summary
This study compared statistical methods for Alzheimer's disease (AD) proteomics, finding partial least squares discriminant analysis superior for cross-species comparisons. Extracellular matrix and integrin pathways were key findings in both human and mouse models.
Area of Science:
- Neuroscience
- Genetics
- Biochemistry
Background:
- Understanding genotype-phenotype relationships is crucial for complex diseases like Alzheimer's disease (AD).
- Proteomics is extensively used to study AD pathogenesis, but cross-species statistical comparisons between human and mouse models are underexplored.
- Challenges in cross-species analysis include collinearity, dimensionality reduction, and protein matching.
Purpose of the Study:
- To evaluate two statistical methods for identifying proteins and pathways in Alzheimer's disease (AD) using cross-species proteomic data.
- To assess the reliability of mouse models in mimicking human AD-related proteomic changes.
- To inform the development of more relevant mouse models for late-onset Alzheimer's disease research.
Main Methods:
- Utilized proteomic data from a well-characterized human cohort (22+ years follow-up).
- Generated proteomic data from whole brains of CVN-AD mouse models and controls.
- Applied and compared LASSO regression and partial least squares discriminant analysis (PLS-DA) for cross-species statistical comparison.
Main Results:
- Partial least squares discriminant analysis (PLS-DA) demonstrated superior statistical performance compared to LASSO regression for proteomics analysis.
- Identified dysregulation of extracellular matrix proteins and integrin-related pathways in both human AD and mouse models.
- The study validated the utility of specific statistical approaches for cross-species proteomic analysis in Alzheimer's disease.
Conclusions:
- Partial least squares discriminant analysis (PLS-DA) is a more effective method for cross-species proteomic comparisons in Alzheimer's disease (AD) research.
- Extracellular matrix and integrin pathways represent significant biological targets in both human AD and relevant mouse models.
- This validated approach enhances the reliability of mouse models for studying human late-onset Alzheimer's disease (AD).
