Pathway analysis of rare variants for the clustered phenotypes by using hierarchical structured components analysis

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Large-scale genetic studies frequently involve clustered phenotypes, such as repeated measurements.
• Analyzing clustered phenotypes offers increased statistical power for genetic association detection compared to univariate analyses.
• Integrating biological information can enhance the detection of weak effects from rare variants.

Purpose of the Study:

• To introduce a novel pathway-level association test for clustered phenotypes.
• To improve the detection of genetic associations by analyzing pathways in a unified model.
• To enhance statistical power in genetic studies involving complex phenotypic data.

Main Methods:

• Combined hierarchical structured components analysis and penalized generalized estimating equations.
• Developed a unified model to analyze all pathways while accounting for their correlations.
• Implemented the method in C/C++ and made it publicly available as PHARAOH-GEE.

Main Results:

• Simulation studies demonstrated superior performance compared to existing pathway-level analysis methods for clustered phenotypes.
• A real data analysis of 995 samples with metabolic syndrome-related phenotypes was successfully conducted.
• Identified the glyoxylate and dicarboxylate metabolism pathway, which was missed by univariate analyses and other methods.

Conclusions:

• The proposed PHARAOH-GEE method provides a powerful approach for pathway-level association testing in clustered phenotypes.
• The method successfully identified a biologically relevant pathway in a large-scale dataset.
• PHARAOH-GEE offers an advancement in analyzing complex genetic data with multiple correlated phenotypes.