SKAT โ Why We Do Not Evaluate Rare Variants Individually
Why this matters
Rare variants are tested in aggregate at the gene or region level due to low individual observation counts. The Burden test is powerful under the assumption that the effects of included variants act in a similar direction, whereas SKAT (Sequence Kernel Association Test) is more flexible for sets where effect directions and magnitudes vary or when non-causal variants are mixed in. SKAT-O combines these two extremes but does not provide an automatic optimal solution. Key components such as the variant set definition, weighting scheme, minor allele frequency (MAF) threshold, phenotype model, ancestry/relatedness structure, and multiple-testing correction plan all influence the results. A significant set does not immediately identify which specific variants are causal or clinically relevant.
Interpretation boundary
This article explains the concept and the limits of the evidence. It does not provide diagnostic, testing, or treatment decisions for an individual patient.