Imputation methods based on the Li and Stephens (2003) model of DNA sequence variation (like IMPUTE2 and MaCH) are well suited to performing imputation in African GWAS. As shown in our HapMap 3 and MalariaGEN cross-validations, this kind of model can consistently produce higher accuracy than clustering approaches like the one used by Beagle, for reasons we discuss below. When the model is implemented via an efficient algorithm like IMPUTE2, this accuracy can be achieved at a fraction of the computational price.
