The initial study population included 2,500 children from the Kombos region of The Gambia. As expected, the authors found high levels of stratification based on the four primary ethnic groups in this population. The GWA study was performed on 958 cases with severe malaria and 1,398 controls, using the Affymetrix 500K platform, which is widely used for studies in other populations. The strongest association signal was found at the
HBB
locus (which encodes the haemoglobin beta chain), the best established genetic susceptibility factor for malaria. Using an additional 1,087 cases and 2,376 controls, the authors report strong replication of the HBB association, and suggestive evidence for associations at
SCO1
and
DDC
. The GWA study did not show significant association for any other previously associated loci; the reasons are unclear but, in addition to questioning whether some of these earlier associations will hold, this may reflect limited power in the current study, differences in allele frequencies between populations and how well the platform used was able to tag the relevant genetic variants. This calls attention to the need to develop SNP arrays that are specifically designed to account for the different patterns of genetic variation in Africa.
To fine map the association at the HBB locus, the authors sequenced 111 kb around the association signal in 62 of the Gambian controls. Using this additional SNP variation to impute genotypes in their initial GWA scan, they pinpoint the known causal variant as the highest association signal, demonstrating how population-based resequencing combined with imputation can be used to boost association signals in GWA studies. This approach, also used in the current 1000 Genomes Project, may prove useful for genetic association studies in African populations. The 1000 Genomes Project is also sequencing the genomes of 400 individuals from Africa, providing a valuable reference panel. As for the MalariaGEN network, they are already gearing up for a larger-scale study across 11 African populations.
This is a preview of subscription content, access via your institution