Rare variant association testing for next-generation sequencing data via hierarchical clustering

Area of Science:

• Genetics
• Statistical methods
• Computational biology

Background:

• Complex diseases often involve genetic susceptibility from low-frequency and rare variants.
• Next-generation sequencing enables rare variant association studies in large populations.
• Developing powerful statistical methods is crucial for detecting these associations.

Purpose of the Study:

• To propose and evaluate a novel hierarchical clustering and similarity kernel-based association test for continuous phenotypes.
• To compare the power of this new method against existing burden tests for rare variant association analysis.

Main Methods:

• A hierarchical clustering approach is used to group genetically similar individuals.
• A similarity kernel-based association test is applied to these clusters for continuous phenotypes.
• The performance is evaluated against collapsing methods and burden tests.

Main Results:

• The proposed method shows comparable power to collapsing methods when causal variants have a consistent effect direction.
• It significantly outperforms burden tests when both risk and protective variants are present.
• The Sequence Kernel Association Test (SKAT) emerges as the most powerful approach across considered allelic architectures.

Conclusions:

• The analytical framework underlying SKAT provides higher statistical power for rare variant association studies.
• SKAT effectively controls type I error rates, ensuring reliable detection of genetic associations.
• This approach offers a robust tool for dissecting the genetic architecture of complex diseases.