BayesKAT: bayesian optimal kernel-based test for genetic association studies reveals joint genetic effects in complex...

Area of Science:

• Genetics
• Statistical genetics
• Computational biology

Background:

• Genome-wide Association Studies (GWAS) identify single nucleotide polymorphisms (SNPs) for phenotypes.
• Complex diseases are often polygenic, involving multiple non-linearly dependent genetic variants.
• Existing kernel-based tests (KBT) face challenges in optimal kernel selection, leading to statistical issues.

Purpose of the Study:

• To develop a novel Bayesian framework, BayesKAT, for robust genetic association testing.
• To overcome limitations of existing KBT methods, including type I error inflation and scalability.
• To adaptively select optimal composite kernels for simultaneous genetic association testing.

Main Methods:

• Developed a Bayesian framework (BayesKAT) for adaptive composite kernel selection.
• Implemented a scalable computational strategy for high-dimensional genetic data.
• Evaluated performance using simulated and large-scale real genetics data.

Main Results:

• BayesKAT adaptively selects optimal kernels, addressing kernel specification issues in KBT.
• The framework demonstrates superior performance in detecting complex group-level genetic associations.
• BayesKAT effectively controls type I errors while maintaining high statistical power.

Conclusions:

• BayesKAT offers a powerful and scalable solution for analyzing complex polygenic diseases.
• The method provides mechanistic insights into human diseases by analyzing biologically relevant genetic variant groups.
• BayesKAT advances the field of statistical genetics for complex trait analysis.