MixedBayes: An R Package for Longitudinal Gene-Environment Interaction Analysis Using Robust Sparse Bayesian Mixed...

Area of Science:

• Genetics
• Biostatistics
• Computational Biology

Background:

• High-dimensional gene-environment interactions in longitudinal studies are complex.
• Existing variable selection methods lack robust inferential tools for quantifying interaction uncertainty.
• Longitudinal data analysis requires methods that handle intra-cluster correlations and repeated measures.

Purpose of the Study:

• Introduce the R package mixedBayes (version 0.2.5) for robust high-dimensional gene-environment interaction analysis in longitudinal studies.
• Implement fully Bayesian robust mixed-effects models to address limitations in current inferential tools.
• Provide a framework for analyzing interactions between omics features, treatment effects, genetic main effects, and environmental factors.

Main Methods:

• Developed the mixedBayes R package implementing two classes of Bayesian robust mixed-effects models.
• Utilized Markov chain Monte Carlo (MCMC) for posterior Bayesian inference.
• Applied the models to analyze longitudinal asthma data with high-dimensional SNP measurements.

Main Results:

• The mixedBayes package enables robust analysis of high-dimensional gene-environment interactions in longitudinal data.
• The implemented models successfully accommodate complex correlation structures and heavy-tailed repeated measures.
• Demonstrated the utility of the package through numerical examples and a real-world asthma case study.

Conclusions:

• The mixedBayes R package offers a valuable tool for robust and statistically sound analysis of gene-environment interactions in longitudinal omics data.
• The Bayesian approach provides a robust framework for quantifying uncertainty in complex interaction models.
• Facilitates advanced genetic association studies and personalized medicine research.