Imputation-based meta-analysis of severe malaria in three African populations

Abstract

Combining data from genome-wide association studies (GWAS) conducted at different locations, using genotype imputation and fixed-effects meta-analysis, has been a powerful approach for dissecting complex disease genetics in populations of European ancestry. Here we investigate the feasibility of applying the same approach in Africa, where genetic diversity, both within and between populations, is far more extensive. We analyse genome-wide data from approximately 5,000 individuals with severe malaria and 7,000 population controls from three different locations in Africa. Our results show that the standard approach is well powered to detect known malaria susceptibility loci when sample sizes are large, and that modern methods for association analysis can control the potential confounding effects of population structure. We show that pattern of association around the haemoglobin S allele differs substantially across populations due to differences in haplotype structure. Motivated by these observations we consider new approaches to association analysis that might prove valuable for multicentre GWAS in Africa: we relax the assumptions of SNP-based fixed effect analysis; we apply Bayesian approaches to allow for heterogeneity in the effect of an allele on risk across studies; and we introduce a region-based test to allow for heterogeneity in the location of causal alleles.

Figure 1. Per-sample imputation accuracy measured by r² between true genotypes and genotypes predicted by imputation, averaged over imputation chunks.

Black vertical line shows typical imputation accuracy in a UK population, taken from . Gambian samples (red) perform worst due to the poor coverage of African variation by the Illumina 550 K platform, followed by Kenyan samples (green) on the Illumina Omni2.5M, which while dense has limited overlap with our HapMap3 reference, with Malawian samples (yellow) performing best. Note that imputation accuracy in the Kenyan sample showed a bimodal distribution driven at least partially by ethnic ancestry (Figures S12 and S13).

Figure 2. Principle components analysis.

Top left: principal components analysis (PCA) of the African populations from Hapmap 3 (LWK = Luhya in Webuye, Kenya, 90 individuals; MKK = Maasai in Kinyawa, Kenya, 143 individuals; YRI = Yoruba in Ibadan, Nigeria, 113 individuals) with 500 randomly selected control samples from each of the three study cohorts. Top right, bottom left, bottom right: PCA of all non-excluded samples in each study cohort, coloured by reported ethnic group. Ethnic group is shown as “OTHER” for groups constituting less than 5% of individuals in the cohort, or where the ethnic group was unreported.

Figure 3. Patterns of association around the HBB and ABO loci.

In each figure the top three panels of the plot is the P value of the logistic regression analysis using 5 PCAs in each cohort with the fixed-effect meta-analysis P value shown in the bottom panel. Circles represent genotyped SNP and triangles imputed SNPs. SNPs in r²>0.2 with the functional SNPs in each region:rs334 in HBB which encodes the sickle locus [HbS], and rs8176719 in ABO are coloured with yellow indicating more correlation and red indicating less. Vertical lines indicate the positions of the functional SNPs.

Figure 4. Comparison of fixed and structure effect Bayes factor at autosomal SNPs.

Red dots indicate SNPs mapping to the HBB region in and blue dots indicate those mapping to the ABO region (see Figure 3). Note that points below the diagonal line have more evidence for association under a structured effects model, whereas those above the line have more evidence for association under a fixed effect model.

Figure 5. Evidence for association at approximately 1.3M autosomal SNPs.

The plot shows the log10 Bayes factor comparing the structured effects model (model 4; see main text) to a model of no association. Chromosomes are coloured alternatively light and dark blue. The horizontal line indicates regions with strong evidence of association (BF>1e4).