A Sequence Kernel Association Test for Dichotomous Traits in Family Samples under a Generalized Linear Mixed Model

Area of Science:

• Genetics
• Statistical genetics
• Bioinformatics

Background:

• Identifying rare variants associated with complex diseases in families is challenging due to limitations in current statistical methods.
• Existing set-based methods may lack the power to detect associations when applied to family-based studies with dichotomous traits.

Purpose of the Study:

• To develop and evaluate a novel set-based statistical method for genetic association studies in family samples with dichotomous traits.
• To address the underpowered nature of existing methods for detecting multiple rare variants in complex diseases within families.

Main Methods:

• Introduced a framework using generalized linear mixed models and kernel machine regression for family-based association testing.
• Proposed F-SKAT (Family-based Sequence Kernel Association Test) as an extension of SKAT and famSKAT for dichotomous traits in family data.
• Utilized simulation studies to compare the performance of F-SKAT against existing methods.

Main Results:

• Original SKAT demonstrated inflated type I error rates when applied directly to family data.
• The proposed F-SKAT method maintained correct type I error rates in simulations.
• F-SKAT consistently showed higher statistical power than SKAT (using unrelated individuals) and another family-based method across all evaluated scenarios.

Conclusions:

• F-SKAT is a robust set-based association test suitable for analyzing family data with dichotomous phenotypes.
• The method effectively handles genetic variants with concordant or discordant effects and accommodates various family structures.
• F-SKAT offers improved power for detecting associations between rare variants and complex diseases in family studies.