Design of DNA pooling to allow incorporation of covariates in rare variants analysis

Background: Rapid advances in next-generation sequencing technologies facilitate genetic association studies of an increasingly wide array of rare variants. To capture the rare or less common variants, a large number of individuals will be needed. However, the cost of a large scale study using whole genome or exome sequencing is still high. DNA pooling can serve as a cost-effective approach, but with a potential limitation that the identity of individual genomes would be lost and therefore individual characteristics and environmental factors could not be adjusted in association analysis, which may result in power loss and a biased estimate of genetic effect. Methods: For case-control studies, we propose a design strategy for pool creation and an analysis strategy that allows covariate adjustment, using multiple imputation technique. Results: Simulations show that our approach can obtain reasonable estimate for genotypic effect with only slight loss of power compared to the much more expensive approach of sequencing individual genomes. Conclusion: Our design and analysis strategies enable more powerful and costeffective sequencing studies of complex diseases, while allowing incorporation of covariate adjustment.