SKAT

SKAT evaluates associations between sets of genetic variants and phenotypic traits by aggregating variant-level score statistics into a sequence kernel association framework to detect effects of rare and common variants on binary and quantitative traits.

Key Features:

• SNP-set aggregation: Aggregates individual score test statistics from single nucleotide polymorphisms (SNPs) within a specified set such as a gene or genomic region to produce SNP-set level results.
• Rare and common variant analysis: Evaluates cumulative effects of both rare and common variants without relying solely on upweighting rare variants.
• Sequence kernel association tests: Implements sequence kernel association tests to model variant-set effects across samples.
• Phenotype support: Applicable to both dichotomous (binary) and quantitative (continuous) traits.
• Covariate adjustment: Adjusts for covariates, including principal components, to account for population stratification.
• Integration of data types: Allows integration of genome-wide association study (GWAS) data with whole-exome sequencing or deep-resequencing data from the same individuals.
• Power and sample size: Supports power and sample size calculations for sequence association study design.
• Computational performance: Maintains computational efficiency suitable for analysis of variant sets.
• Validation and benchmarking: Validated through simulations under diverse scenarios and compared against burden and variance-component tests.

Scientific Applications:

• Disease association studies: Detects associations between sets of variants and disease susceptibility, exemplified by analyses of Crohn disease and autism spectrum disorders.
• Integrative genetic analysis: Combines GWAS genotypes with whole-exome or deep-resequencing data to assess locus-level allelic spectra.
• Study design: Informs power and sample size decisions for sequencing association studies.
• Trait analysis: Applied to both case-control (binary) and quantitative trait analyses.

Methodology:

Aggregates individual SNP score statistics within specified sets and computes SNP-set level p-values using sequence kernel association tests, with covariate adjustment (including principal components), validated by simulation and supporting power/sample size calculations.