A unified method for detecting secondary trait associations with rare variants: application to sequence data

Area of Science:

• Genetics
• Biostatistics
• Bioinformatics

Background:

• Next-generation sequencing enables rare variant (RV) detection for quantitative traits (QT).
• Studies often select samples with extreme QT due to high sequencing costs, leading to underpowered individual analyses.
• Secondary traits are often measured but analyzing them in selected samples can introduce bias if ascertainment is not modeled.

Purpose of the Study:

• To develop a unified method for analyzing secondary trait associations with RVs in selected samples.
• To overcome limitations of existing methods, including analytical p-value approximations and inability to incorporate powerful RV tests.
• To enable joint analysis of multiple cohorts with different ascertainment designs to boost statistical power.

Main Methods:

• Developed the STAR (Secondary Trait Association with Rare variants) method.
• Incorporated all existing RV association tests, including burden tests, sequence kernel association tests, and variable selection-based methods.
• Enabled statistical significance evaluation through permutations or analytical methods.

Main Results:

• STAR was evaluated using comprehensive simulation studies, demonstrating its performance compared to commonly used RV association tests.
• Application to the SardiNIA project dataset identified a significant association between LDLR and systolic blood pressure.
• The identified association is supported by existing pharmacogenetic studies.

Conclusions:

• STAR is a powerful and flexible tool for detecting RV associations with secondary traits in sequencing studies.
• It overcomes limitations of existing methods by allowing joint analysis of multiple cohorts and incorporating a wider range of RV tests.
• STAR facilitates more robust and powerful genetic association studies for complex diseases.