Boosting the Power of the Sequence Kernel Association Test by Properly Estimating Its Null Distribution

Area of Science:

• Genetics
• Statistical genetics
• Bioinformatics

Background:

• Rare-variant association studies are crucial for identifying genetic variants linked to diseases.
• The Sequence Kernel Association Test (SKAT) is a widely used statistical method in this field.
• Current SKAT methods face limitations in statistical power due to using all subjects for parameter estimation.

Purpose of the Study:

• To develop an improved estimation method for the null distribution in rare-variant association studies.
• To enhance the statistical power of the Sequence Kernel Association Test (SKAT) while preserving type I error control.
• To introduce SKAT+, a novel method utilizing control subjects for estimation.

Main Methods:

• Developed SKAT+ (Sequence Kernel Association Test plus), a novel estimation method for null distribution.
• Employed a test statistic identical to SKAT but with a modified estimation approach using only control subjects.
• Validated SKAT+ through extensive simulation studies and real-world data applications (Genetic Analysis Workshop 17, Ocular Hypertension Treatment Study).

Main Results:

• SKAT+ demonstrated superior statistical power compared to the standard SKAT method.
• The method effectively maintained control over the type I error rate.
• Applications to real genetic datasets confirmed the enhanced performance of SKAT+.

Conclusions:

• SKAT+ offers a more powerful approach for rare-variant association studies than traditional SKAT.
• The method's ability to use control subjects for estimation is key to its improved performance.
• SKAT+ is applicable to various extensions of SKAT in genetic research.