Principal interactions analysis for repeated measures data: application to gene-gene and gene-environment...

Area of Science:

• Genetics and Biostatistics
• Longitudinal Data Analysis
• Environmental Health Research

Background:

• Longitudinal cohorts increasingly collect genetic data alongside environmental and lifestyle factors.
• Modeling gene-gene and gene-environment interactions with repeated measures is crucial for understanding complex traits.
• Classical interaction models can fail when interaction structures are misspecified.

Purpose of the Study:

• To propose a robust method for modeling gene-gene and gene-environment interactions using longitudinal data.
• To introduce 'principal interactions analysis' based on singular value decomposition for improved interaction modeling.
• To compare the proposed methods with classical alternatives using real and simulated data.

Main Methods:

• Review of classical Tukey and Mandel models for interaction detection.
• Exploration of additive main effects and multiplicative interaction (AMMI) models via singular value decomposition.
• Introduction of resampling-based methods for unbalanced and longitudinal data.
• Application to data from the Normative Aging Study.

Main Results:

• Principal interactions analysis offers a more robust approach to modeling complex interactions compared to classical methods.
• The method provides useful summaries of subject-specific and time-varying effects.
• Geometric representation of interaction structures is facilitated.
• Simulation studies demonstrate the properties of principal interactions analysis.

Conclusions:

• Principal interactions analysis is a valuable tool for dissecting gene-gene and gene-environment interactions in longitudinal studies.
• The proposed methods enhance the understanding of genetic and environmental influences on health outcomes.
• This approach is applicable to various complex biological and health-related datasets.