Table 2.

Comparison of power to detect disease-SNP association: tagSNPs only; different sequencing designs with fixed sequencing cost; and different sequencing designs with fixed sequencing cost plus imputation in additional individuals not sequenced

For sequencing and imputation into remaining samples, we removed SNPs with estimated r² < 0.3. Association analysis was performed using logistic regression separately on allele dosages for sequenced individuals and imputed individuals. Results were subsequently meta-analyzed using METAL (Willer et al. 2010).