Localizing components of shared transethnic genetic architecture of complex traits from GWAS summary data

Abstract

1Despite strong transethnic genetic correlations reported in the literature for many complex traits, the non-2 transferability of polygenic risk scores across populations suggests the presence of population-specific 3 components of genetic architecture. We propose an approach that models GWAS summary data for one 4 trait in two populations to estimate genome-wide proportions of population-specific/shared causal SNPs. 5In simulations across various genetic architectures, we show that our approach yields approximately 6 unbiased estimates with in-sample LD and slight upward-bias with out-of-sample LD. We analyze 9 7 complex traits in individuals of East Asian and European ancestry, restricting to common SNPs (MAF > 8 5%), and find that most common causal SNPs are shared by both populations. Using the genome-wide 9 estimates as priors in an empirical Bayes framework, we perform fine-mapping and observe that high-10posterior SNPs (for both the population-specific and shared causal configurations) have highly correlated 11 effects in East Asians and Europeans. In population-specific GWAS risk regions, we observe a 2.8x 12 enrichment of shared high-posterior SNPs, suggesting that population-specific GWAS risk regions harbor 13 shared causal SNPs that are undetected in the other GWAS due to differences in LD, allele frequencies, 14 and/or sample size. Finally, we report enrichments of shared high-posterior SNPs in 53 tissue-specific 15 functional categories and find evidence that SNP-heritability enrichments are driven largely by many low-16 effect common SNPs. 17 respect to the set of variants included in the GWAS and can contain variants with indirect nonzero effects 52 that are statistical rather than biological in nature (this is analogous to the definition of SNP-heritability, 53 which is also a function of a specific set of SNPs 11,54-56 ). Through extensive simulations, we show that our 54 method yields approximately unbiased estimates of the proportions of population-specific/shared causal 55 variants if in-sample LD is used and slightly upward-biased estimates if LD is estimated from an external 56 reference panel. We then show that using these estimates as priors to perform fine-mapping (Methods) 57 produces well-calibrated per-SNP posterior probabilities and enrichment test statistics. We note that the 58 definition of enrichment used here is related to, but conceptually distinct from, definitions of SNP-59 heritability enrichment 13,16 . Under our framework, an enrichment of causal SNPs greater than 1 indicates 60 that, compared to the genome-wide background, there are more causal SNPs in that region than expected 57,58 61 (Methods). In contrast, an enrichment of SNP-heritability greater than 1 indicates that the average per-SNP 62 effect size in the region is larger than the genome-wide average per-SNP effect size. 63We apply our approach to publicly available GWAS summary statistics for 9 complex traits and 64 diseases in individuals of East Asian (EAS) and European (EUR) ancestry (average NEAS = 94,621, N...

Analysis across Taiwan Biobank, Biobank Japan and UK Biobank identifies hundreds of novel loci for 36 quantitative traits

“…Prior studies have shown that complex traits and diseases are genetically correlated at different levels between EAS and European (EUR) populations 21–23 , but the genetic overlap within EAS populations has not been characterized yet. Leveraging existing GWAS summary statistics from BBJ and UKBB, we investigated the comparative genetic architecture of quantitative traits within EAS (TWB vs. BBJ), and between EAS and EUR populations (TWB vs. UKBB and BBJ vs. UKBB).…”

Gene expression in African Americans and Latinos reveals ancestry-specific patterns of genetic architecture

“…Tier 2 anc-eQTLs also had nonoverlapping 95% credible sets for each gene but were commonly segregating (MAF ≥ 0.01) in both high and low ancestry groups. This class of anc-eQTLs was identified using PESCA 17 , which accounted for population-specific LD patterns. We identified 109 genes (1.1%) with eQTLs having a posterior probability (PP) > 80% of being specific to AFRhigh and 33 genes (0.4%) matching the same criteria for IAMhigh.…”

“…For genes with non-overlapping lead eQTL signals (Figure 3a, left branch), eQTLs from the high group were categorized into Tier 1 anc-eQTLs if they were common variants (MAF >= 0.01) in the high group but rare or absent in the low group. If the eQTLs were common variants in both the high and low groups, their specificity was further assessed using PESCA 17 for differential effect size while accounting for LD between eQTLs. LD pruning on eQTLs with r2 > 0.95 was performed prior to PESCA analysis.…”

Improving Polygenic Prediction in Ancestrally Diverse Populations

“…Initial studies have indicated that the genetic architectures of many complex traits and diseases are concordant between populations -both at the single-variant level and at the genome-wide level. In particular, many genetic associations identified in European populations have been replicated in non-European GWAS, and recent studies have estimated the trans-ethnic genetic correlation between European and non-European populations to be moderate or high for many traits [16][17][18][19] . These findings suggest that the transferability of PRS may be improved by integrating GWAS summary statistics from diverse populations.…”