On Sample Size and Power Calculation for Variant Set-Based Association Tests.

Area of Science:

• Genetics and Genomics
• Statistical Genetics
• Bioinformatics

Background:

• Sample size and power calculations are critical for designing robust sequence-based association studies.
• Current methods, like SEQPower and SPS, rely on computationally intensive Monte Carlo simulations for power estimation.
• Existing analytical approaches for SKAT power may overestimate results, particularly at small significance levels relevant for large-scale sequencing projects.

Purpose of the Study:

• To develop computationally efficient and accurate methods for sample size and power calculations in sequence-based association studies.
• To address the limitations of existing analytical methods for SKAT power estimation.
• To provide reliable tools for planning large-scale whole genome and exome sequencing projects.

Main Methods:

• Proposed a novel chi-squared (χ(2)) approximation-based approach for efficient and accurate power computation.
• Developed and implemented a more accurate "exact" method for power calculation, offering improved precision over Monte Carlo simulations.
• Validated the proposed methods against existing approaches and Monte Carlo simulations.

Main Results:

• The proposed chi-squared approximation method provides accurate and efficient power calculations.
• The "exact" method delivers highly accurate power estimates and can serve as a benchmark for other approximation techniques.
• Both new methods are more efficient than traditional Monte Carlo simulations for power estimation.

Conclusions:

• The developed chi-squared approximation and "exact" methods offer significant improvements in accuracy and efficiency for SKAT power calculations.
• These methods are crucial for optimizing the design of sequence-based association studies, especially for whole genome and exome sequencing.
• Publicly available R programs implementing these methods facilitate their adoption in genetic research planning.