Power and sample size calculations for designing rare variant sequenc - ing association studies

Recently, Wu et al. [4] have proposed the sequence kernel machine test (SKAT) to test association between genetic variants in a gene or region and a continuous or binary trait. SKAT, which uses the kernel machine regression framework, is very flexible and computationally efficient. From extensive simulation studies and real data application, it has been shown that SKAT is more powerful than the collapsing based burden tests under many circumstances [4]. To design new sequence association study with SKAT as a testing procedure, it is important to know the required sample size to achieve a proper statistical power. Power and sample size calculation can be done by simulations, however this computer intensive approach would be time consuming. Here, we derive the analytical formula of the statistical power of SKAT. Required sample size can be computed easily by inverting the power function. In addition, We have developed user friendly R package which implements the power and sample size calculation formula (Web Resources).