UKB

UKB implements generalized linear mixed model (GLMM)-based genome-wide association analysis for binary traits in biobank-scale datasets.

Key Features:

• Efficiency and Speed: Leverages sparse matrix-based algorithms to achieve severalfold to orders-of-magnitude speed improvements over existing state-of-the-art tools for large-scale datasets.
• Scalability: Processes cohorts with millions of individuals and large numbers of variants and traits without compromising performance.
• Statistical Calibration: Demonstrates well-calibrated test statistics for both common and rare variants under the null, including scenarios with extreme case-control ratios, based on simulation studies.
• Application to Rare Variants: Utilizes imputed genotype data from large cohorts to detect associations between rare variants and binary complex traits.

Scientific Applications:

• Binary Trait Analysis: Performs GWAS for binary traits using GLMM-based inference to improve statistical properties relative to traditional linear mixed model-based approaches.
• Large Cohort Studies: Enables large-scale genetic association studies on datasets containing millions of individuals and numerous variants.
• Rare Variant Discovery: Applied to UK Biobank imputed data to identify 259 rare variants associated with 75 binary traits.

Methodology:

Implements generalized linear mixed models (GLMMs) using sparse matrix-based algorithms, evaluates calibration via simulation studies, and operates on imputed genotype data for rare-variant association testing.