A Set-Based Mixed Effect Model for Gene-Environment Interaction and Its Application to Neuroimaging Phenotypes.

Area of Science:

• Neuroscience
• Genetics
• Biostatistics

Background:

• Imaging genetics links genetic variation to brain mechanisms but faces statistical challenges with high-dimensional data.
• Studying gene-environment interactions (G×E) on neuroimaging phenotypes is particularly complex.

Purpose of the Study:

• To propose a novel set-based mixed effect model (MixGE) for analyzing G×E on neuroimaging phenotypes.
• To enhance statistical power and computational efficiency in imaging genetics studies.

Main Methods:

• Developed MixGE incorporating fixed and random effects for G×E analysis.
• Utilized score statistics to avoid computationally intensive parameter estimation.
• Combined score statistics for increased statistical power.

Main Results:

• MixGE demonstrated superior statistical power compared to existing methods using simulated and Alzheimer's Disease Neuroimaging Initiative (ADNI) data.
• The model's computational efficiency enabled voxelwise G×E analysis on tensor-based morphometry (TBM).
• Identified a potential G×E between the ABCA7 gene and cardiovascular risk affecting right superior parietal cortex volume.

Conclusions:

• MixGE offers a powerful and computationally efficient approach for G×E research in imaging genetics.
• The findings highlight a specific gene-environment interaction relevant to neuroimaging phenotypes, warranting further investigation.