Kernel-Based Aggregation of Marker-Level Genetic Association Tests Involving Copy-Number Variation

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Genetic association studies often face challenges with copy-number variants (CNVs) due to their extended nature across multiple markers.
• Individual marker tests for CNVs typically have low statistical power.
• Pooling information across markers within a CNV is desirable but complicated by unknown CNV boundaries.

Purpose of the Study:

• To propose and evaluate a kernel-based method for aggregating marker-level genetic association tests for CNVs.
• To investigate methods for pooling information across markers spanned by CNVs when boundaries are unknown.
• To develop a statistically rigorous approach for CNV association testing that maintains family-wise error rate.

Main Methods:

• A kernel-based method is proposed for aggregating association test results from multiple markers within a CNV.
• A permutation-based approach is developed to preserve the family-wise error rate (FWER) in CNV association testing.
• Simulations based on real pharmacogenomic data were used to assess the empirical power.

Main Results:

• The proposed kernel-based aggregation method demonstrates favorable empirical power in simulations.
• The permutation-based approach effectively controls the family-wise error rate, unlike simpler alternatives.
• The method shows advantages in analyzing pharmacogenomic data, such as for gemcitabine response.

Conclusions:

• The kernel-based aggregation method offers a powerful and statistically sound approach for genetic association testing of CNVs.
• This method addresses the challenge of unknown CNV boundaries by effectively pooling marker information.
• The findings have implications for identifying genetic variants, including CNVs, associated with drug response and other complex traits.