An adaptive combination method for Cauchy variable based on optimal threshold.

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Set-based analysis is popular for rare variant association studies but sensitive to genetic architecture.
• Directly combining association signals can introduce noise from non-causal variants.

Purpose of the Study:

• To develop an adaptive Cauchy-variable combination (ACC) method to improve rare variant association analysis.
• To address the sensitivity of existing methods to genetic architecture and noise.

Main Methods:

• Extended the aggregated Cauchy association test (ACAT).
• Proposed ACC, which combines Cauchy-transformed P-values and adaptively truncates noise.
• ACC utilizes summary statistics, enabling analysis without original genotype/phenotype data.

Main Results:

• ACC demonstrated higher statistical power than comparative methods when few variants are causal.
• The method proved robust across varied genetic architectures.
• ACC showed strong performance in simulation studies and real data analysis (GAW19).

Conclusions:

• ACC offers a powerful and robust approach for rare variant association studies.
• The adaptive noise truncation and use of summary statistics enhance its applicability.
• ACC is a valuable tool for genetic architecture analysis and disease association.