fastGWA-GLMM

fastGWA-GLMM performs genome-wide association testing using generalized linear mixed models (GLMMs) to enable scalable, statistically robust association analysis of binary and other traits in biobank-scale datasets.

Key Features:

• Computational Efficiency: Employs sparse matrix-based algorithms to optimize computation time and memory, achieving approximately 37× speed improvements on datasets up to 400,000 individuals.
• Statistical Robustness: Produces well-calibrated test statistics under the null hypothesis and maintains accuracy for extreme case–control ratios (e.g., 0.1%) across both common and rare variants.
• Application to Large Cohorts: Successfully applied to the UK Biobank (456,348 individuals, 11,842,647 variants, 2,989 binary traits), identifying 259 rare variants associated with 75 traits.

Scientific Applications:

• Binary Trait Analysis: Enables association testing for binary traits where linear mixed models may be suboptimal, retaining calibration at extreme case–control ratios.
• Rare Variant Discovery: Detects rare variant associations in large cohorts, as demonstrated by the identification of 259 rare variants linked to 75 traits in the UK Biobank.

Methodology:

Fits generalized linear mixed models and implements sparse matrix-based algorithms for scalable computation, with test statistics evaluated for calibration under the null hypothesis.