Powerful Genetic Association Analysis for Common or Rare Variants with High-Dimensional Structured Traits

Area of Science:

• Genetics
• Statistical Genomics
• Bioinformatics

Background:

• Genetic association studies often collect data on multiple complex traits.
• Correlated traits may share genetic mechanisms, making joint analysis more powerful and interpretable.
• Existing methods struggle with high-dimensional and structured traits like pathway gene expressions.

Purpose of the Study:

• To propose a novel statistical test, the dual kernel-based association test (DKAT), for analyzing associations between multiple traits and genetic variants.
• To address limitations of existing methods in handling high-dimensional and structured phenotypic data.
• To enable joint analysis of common and rare genetic variants across multiple complex traits.

Main Methods:

• Developed the dual kernel-based association test (DKAT).
• DKAT utilizes two kernels to capture phenotypic and genotypic similarity between subjects.
• Employs an analytical P-value calculation, avoiding computationally intensive resampling.

Main Results:

• DKAT demonstrates correct type-I error rates in simulations.
• The method shows higher statistical power compared to existing approaches.
• Successful application to analyze the genetic regulation of pathway gene expressions.

Conclusions:

• DKAT is a robust and powerful method for testing associations between multiple traits and genetic variants.
• The kernel-based approach effectively handles high-dimensional and structured data.
• DKAT provides a statistically sound and computationally efficient alternative for genetic association studies.