Fig. 1

Experimental steps for creating the data sets in the pooling and classical imputation scenarios.The original data set “Chr20 x OmniExpress” consists of the genotype data of 2504 samples at 52,697 SNPs. The set of markers is created by intersecting the variants present on both bead-chips from the Illumina manufacturer and the data for the chromosome 20 in the 1KGP. The original data set is randomly split into a reference panel and a study population. In the LDHD scenario, all markers in the HD data set that are not present in the LD data set are filtered out in the study population. In the pooled HD scenario, the study samples are first assigned to blocks and pools, second the pools are genotyped at all markers in the HD data set, and last the genotype of each sample is decoded from the pools at every marker. See Fig. 3 for an example of the simulation steps in 1 block at 1 marker. The imputation step is performed in both scenarios from the reference panel, with Beagle on the one hand and Prophaser on the other hand. The genotyping accuracy in each scenario is computed by comparing the imputed genotypes with the true ones in the original data set for the study population