Trans-ethnic study design approaches for fine-mapping

Abstract

Studies that traverse ancestrally diverse populations may increase power to detect novel loci and improve fine-mapping resolution of causal variants by leveraging linkage disequilibrium differences between ethnic groups. The inclusion of African ancestry samples may yield further improvements because of low linkage disequilibrium and high genetic heterogeneity. We investigate the fine-mapping resolution of trans-ethnic fixed-effects meta-analysis for five type II diabetes loci, under various settings of ancestral composition (European, East Asian, African), allelic heterogeneity, and causal variant minor allele frequency. In particular, three settings of ancestral composition were compared: (1) single ancestry (European), (2) moderate ancestral diversity (European and East Asian), and (3) high ancestral diversity (European, East Asian, and African). Our simulations suggest that the European/Asian and European ancestry-only meta-analyses consistently attain similar fine-mapping resolution. The inclusion of African ancestry samples in the meta-analysis leads to a marked improvement in fine-mapping resolution.

Figure 1

Boxplots, for perfect data, of the number of SNPs in the 95% credible sets for summary measures of the five loci and for individual loci at MAFs 5, 10, and 20% and ancestral diversities single, moderate, and high.

Figure 2

Boxplots, for imputed data, of the number of SNPs in the 95% credible sets for summary measures of the five loci and for individual loci at MAFs 5, 10, and 20% and ancestral diversities single, moderate, and high.