Looking ahead, we anticipate that many of the initial African GWAS will be conducted in west African populations; for example, such populations constitute a large part of MalariaGEN's Consortial Project 1. The power of these studies could potentially be boosted by using African American haplotypes to augment the reference sets collected in Africa. While there are clear merits to this idea, one might worry that the haplotype segments of non-African ancestry would pose problems for imputation. We can address this question by inspecting the IMPUTE2 results from our HapMap 3 cross-validations. Encouragingly, these results show that adding the HapMap 3 ASW panel to the reference set improved accuracy in every African cross-validation panel (File S1).
