UK Biobank Whole-Exome Sequence Binary Phenome Analysis with Robust Region-Based Rare-Variant Test

Area of Science:

• Genetics
• Bioinformatics
• Statistical Genetics

Background:

• Binary phenotypes in biobank data often exhibit unbalanced case-control ratios, leading to inflated Type I error rates in association tests.
• Existing region-based tests struggle with extreme imbalance or scalability for large datasets.

Purpose of the Study:

• To develop accurate and scalable region-based tests for genetic association analysis with unbalanced binary phenotypes.
• To address limitations of current methods in controlling Type I error rates under extreme case-control imbalance.

Main Methods:

• Proposed SKAT- and SKAT-O- type region-based tests calibrating single-variant score statistics using saddle point approximation (SPA) and efficient resampling (ER).
• Simulation studies to evaluate Type I error rates and calibration of p-values.
• Application to UK Biobank whole-exome sequence data (45,596 samples, 791 phenotypes).

Main Results:

• The proposed method demonstrated well-calibrated p-values in simulations, contrasting with greatly inflated Type I error rates (90x exome-wide alpha) of unadjusted approaches at a 1:99 ratio.
• The method showed similar computational time to unadjusted approaches, confirming scalability for large-scale biobank data.
• Analysis of UK Biobank data identified 10 rare-variant associations (p < 10^-7), including JAK2/myeloproliferative disease, HOXB13/prostate cancer, and F11/congenital coagulation defects.

Conclusions:

• The novel SPA and ER-calibrated region-based tests provide accurate and scalable solutions for analyzing unbalanced biobank data.
• These methods enhance the identification of rare-variant associations with complex diseases.
• Publicly available results via a web server facilitate further genetic research.