Multiple comparisons in genetic association studies: a hierarchical modeling approach.

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Multiple testing poses a significant challenge in genetic association studies with numerous variants.
• Existing methods struggle to efficiently handle the high dimensionality of genetic data.

Purpose of the Study:

• To propose a novel hierarchical modeling approach to mitigate the multiple comparisons problem in genetic association studies.
• To develop a method for computing the effective number of genetic effects and a new adjustment for multiple comparisons.

Main Methods:

• Simultaneously fitting multiple variables using hierarchical models, shrinking unimportant effects towards zero.
• Developing a method for effective number of genetic effects in hierarchical generalized linear models.
• Proposing a hierarchical Bonferroni correction based on the effective number of genetic effects.

Main Results:

• Hierarchical models provide more efficient parameter estimates compared to traditional methods.
• The approach effectively reduces the number of "significant" effects, addressing multiple comparisons.
• Increased power to detect disease-associated variants and controlled Type I error rates were demonstrated.

Conclusions:

• The proposed hierarchical modeling approach offers an effective solution for multiple comparisons in genetic association studies.
• The method enhances statistical power and maintains Type I error control.
• A freely available R package, BhGLM, implements this novel approach.