Modeling genotype-protein interaction and correlation for Alzheimer's disease: a multi-omics imaging genetics study

Area of Science:

• Neuroscience
• Genetics
• Biostatistics

Background:

• Alzheimer's disease (AD) research integrates genomics, proteomics, and radiomics for comprehensive understanding.
• Current methods often neglect non-additive effects like genotype-protein interactions (GPI) and correlation patterns in brain imaging genetics.
• These overlooked non-additive effects may significantly influence intermediate imaging phenotypes and disease progression.

Purpose of the Study:

• To address the unexploited challenge of disentangling genotype-protein interactions (GPI) from correlation patterns in brain imaging genetics.
• To introduce a novel computational method, Multi-Task Genotype-Protein Interaction and Correlation (MT-GPIC), for identifying GPI and extracting correlation patterns.
• To enhance the interpretability and stability of findings by utilizing novel penalties and considering brain region interconnectedness.

Main Methods:

• Development of the MT-GPIC method to simultaneously identify genotype-protein interactions and correlation patterns.
• Application of novel and off-the-shelf penalties for identifying meaningful genetic risk factors and exploiting brain region interconnectedness.
• Implementation of a fast, convergent optimization strategy to address the computational burden of calculating GPI.

Main Results:

• MT-GPIC demonstrated superior performance compared to state-of-the-art methods on the Alzheimer's Disease Neuroimaging Initiative dataset.
• The method achieved higher correlation coefficients and improved classification accuracy in AD-related analyses.
• MT-GPIC successfully identified interpretable, phenotype-related GPI and correlation patterns within high-dimensional omics data.

Conclusions:

• The developed MT-GPIC method effectively disentangles genotype-protein interactions and correlation patterns in AD research.
• Findings suggest that non-additive genetic and protein effects play a crucial role in AD pathogenesis and imaging phenotypes.
• This approach offers enhanced diagnostic accuracy and provides valuable insights into the underlying mechanisms of Alzheimer's disease.