As in the HapMap 3 comparisons, our main goal in designing a cross-validation experiment was to provide guidance on how to use existing and future imputation reference panels. We chose to focus on the GMB panel because genotype imputation has previously been evaluated in a Gambian GWAS (Jallow et al. 2009) and because the 1000 Genomes Project is planning to sequence a set of individuals from the Gambia.
