CauchyGM

CauchyGM performs robust SNP-set association testing by combining score tests, SKAT, and the Cauchy Combination Test to detect genetic associations in GWAS without prespecified inheritance models.

Key Features:

• Model inference without prior knowledge: Uses a generalized linear model framework to infer underlying genetic models within an SNP-set without requiring predefined inheritance patterns.
• Score test with P-value aggregation: Performs a score test that estimates individual P-values for SNPs with minor allele frequency (MAF) > 1% and three genotypes and aggregates these P-values using the Sequence Kernel Association Test (SKAT).
• Cauchy Combination Test: Combines correlated P-values across multiple SNPs and different genetic models within an SNP-set using the Cauchy Combination Test to improve detection under varying genetic architectures.
• Omnibus association test (CauchyGM-O): Implements CauchyGM-O to combine CauchyGM with SKAT and the burden test to accommodate both sparse and dense signal patterns.
• Type I error control and power improvement: Simulation results demonstrate maintenance of type I error rates at the genome-wide significance level and substantial power gains compared to existing methods.

Scientific Applications:

• Pharmacogenomic GWAS analysis: Applied to data from a large cardiovascular randomized clinical trial to identify novel genome-wide significant genes and uncover genetic associations relevant to complex traits and personalized medicine.

Methodology:

SNP-set level analysis of SNPs with MAF > 1% and three genotypes using a generalized linear model framework; score test to obtain individual SNP P-values; P-value aggregation with SKAT; combination of correlated P-values via the Cauchy Combination Test; omnibus approach (CauchyGM-O) combining CauchyGM, SKAT, and the burden test.