A mixture model approach in gene-gene and gene-environmental interactions for binary phenotypes

Area of Science:

• Genetics
• Biostatistics
• Translational Research

Background:

• Genetic association studies aim to link genetic/environmental factors to clinical phenotypes.
• Predicting phenotypes based on genetic and environmental combinations aids clinical decision-making.
• Traditional logistic regression struggles to definitively group genotype combinations for phenotype prediction.

Purpose of the Study:

• To develop a mixture model for detecting gene-environment interactions and clustering phenotypes.
• To provide a robust statistical method for translational research in genetic association studies.
• To enhance clinical decision-making by accurately grouping genetic and environmental factors.

Main Methods:

• A novel mixture model approach was proposed, utilizing a likelihood ratio test (LRT).
• The model concurrently detects main and interaction effects of genetic and environmental variables.
• Phenotype cluster analysis is performed based on identified genetic and environmental variable combinations.

Main Results:

• The mixture model's LRT demonstrated robustness to small and unequal sample sizes.
• Extensive simulations confirmed the mixture model's superior power in detecting multi-way gene/environmental combinations.
• The mixture model exhibited the highest probability of recovering true partitions in simulations.

Conclusions:

• The proposed mixture model offers a powerful and robust method for genetic association studies.
• This approach effectively addresses the challenge of grouping genetic and environmental factors for phenotype prediction.
• The method has practical applications, as demonstrated in interim analyses of cancer studies.