Adaptive tests for detecting gene-gene and gene-environment interactions

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Detecting gene-gene and gene-environment interactions is crucial in genetic association studies.
• High-dimensional data presents statistical challenges, including reduced power and multiple testing burdens.
• Existing methods often involve dimension reduction or specialized high-dimensional tests.

Purpose of the Study:

• To propose and evaluate adaptive statistical tests for detecting gene-gene and gene-environment interactions.
• To combine dimension reduction with robust high-dimensional statistical testing.
• To improve statistical power in genetic association studies with complex interaction effects.

Main Methods:

• Extension of the adaptive Neyman test for high-dimensional data.
• Application of adaptive sum of squared score (SSU) and weighted sum of squared score (SSUw) tests.
• Comparison of adaptive tests with their non-adaptive versions on simulated genetic data.

Main Results:

• Adaptive SSU and SSUw tests demonstrated improved power over non-adaptive versions for sparse interaction models.
• The proposed adaptive tests performed consistently well across various simulation scenarios.
• These methods are applicable within a general regression analysis framework for diverse traits and covariates.

Conclusions:

• Adaptive statistical tests offer a powerful and effective approach for detecting complex genetic interactions.
• The proposed methods provide a valuable tool for analyzing high-dimensional genetic association data.
• These adaptive tests can be broadly applied in genetic research involving various traits and covariates.