LDAK-GBAT

LDAK-GBAT performs gene-based association testing using GWAS summary statistics to identify genes associated with human phenotypes.

Key Features:

• Computational efficiency: Processes imputed data of approximately 2.9 million common, genic SNPs in about 30 minutes with memory usage under 10 gigabytes.
• Uses GWAS summary statistics: Conducts gene-based association tests using summary statistics rather than individual-level genotype data.
• Calibrated p-values: Produces well-calibrated p-values for robust statistical inference.
• Type I error control: Controls type I error rates when an appropriate reference panel is used.
• Increased statistical power: Identifies on average 19% more significant genes than sumFREGAT-ACAT (standard error 1%).
• Benchmarking: Performance compared against sumFREGAT-ACAT, MAGMA, GCTA-fastBAT, sumFREGAT-SKAT-O, and sumFREGAT-PCA.

Scientific Applications:

• Gene discovery in GWAS: Applied to identify gene-level associations across diverse human phenotypes using summary statistics.
• Cross-cohort evaluation: Evaluated on 109 phenotypes from the UK Biobank (82), the Million Veteran Program (18), and the Psychiatric Genetics Consortium (9).

Methodology:

Performs gene-based association tests by leveraging GWAS summary statistics, evaluates p-value calibration and type I error using a reference panel, and was benchmarked against other gene-based methods.