Joint testing of rare variant burden scores using non-negative least squares

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Gene-based burden tests are crucial for exome-wide association studies (EWAS).
• Current methods combine variants into single scores but face challenges with correlated tests and complex multiple testing corrections.
• Interpreting results from multiple correlated burden tests can be difficult.

Purpose of the Study:

• To introduce a novel gene-based burden test method, the sparse burden association test (SBAT).
• To address limitations in existing gene-based association tests, particularly concerning correlated burden scores.
• To provide a robust method for analyzing EWAS data that simultaneously assesses model significance and selects relevant burden scores.

Main Methods:

• SBAT tests a joint set of burden scores, assuming causal scores share the same effect direction.
• The method integrates model significance assessment and burden score selection.
• Implementation within the REGENIE software facilitates its application.

Main Results:

• Simulations demonstrate SBAT's good calibration and superior performance in specific scenarios compared to existing gene-based tests.
• Application to 73 UK Biobank quantitative traits confirms SBAT's value as an additional gene-based testing approach.
• SBAT effectively handles correlated burden scores, simplifying interpretation and multiple testing correction.

Conclusions:

• SBAT offers a statistically robust and computationally efficient approach for gene-based association testing in EWAS.
• The method enhances the analysis of complex genetic data by jointly modeling burden scores.
• SBAT represents a valuable addition to the toolkit for genetic researchers, particularly when used alongside other established methods.