Fine-Mapping and Credible Set Construction using a Multi-population Joint Analysis of Marginal Summary Statistics from Genome-wide Association Studies

Abstract

Recent advancement in Genome-wide Association Studies (GWAS) comes from not only increasingly larger sample sizes but also the shifted focus towards underrepresented populations. Multi-population GWAS may increase power to detect novel risk variants and improve fine-mapping resolution by leveraging evidence from diverse populations and accounting for the difference in linkage disequilibrium (LD) across ethnic groups. Here, we expand upon our previous approach for singlepopulation fine-mapping through Joint Analysis of Marginal SNP Effects (JAM) to a multi-population analysis (mJAM). Under the assumption that true causal variants are common across studies, we implement a novel version of JAM that conditions on multiple SNPs while explicitly incorporating the different LD structures across populations. The mJAM framework can be used to first select index variants using the mJAM likelihood with any feature selection approach. In addition, we present a novel approach leveraging the ideas of mediation to construct credible sets for these index variants. Construction of such credible sets can be performed given any existing index variants. We illustrate the implementation of the mJAM likelihood through two implementations: mJAM-SuSiE (a Bayesian approach) and mJAM-Forward selection. Through simulation studies based on realistic effect sizes and levels of LD, we demonstrated that mJAM performs better than other existing multi-ethnic methods for constructing concise credible sets that include the underlying causal variants. In real data examples taken from the most recent multi-population prostate cancer GWAS, we showed several practical advantages of mJAM over other existing methods.

Genome-wide association study of prostate-specific antigen levels in 392,522 men identifies new loci and improves cross-ancestry prediction

“…In our discovery cohorts, we identified 318 independent genome-wide significant variants in a multi-ancestry analysis of log-transformed PSA levels (Circos plot, Figure 2 ; overall and ancestry-specific Manhattan plots, Figure S1 ; numerical results, Table S4 ; ancestry-specific lead variants Table S5 ) that used multiple reference panels to account for different ancestries (see Methods ). Among them, 184 independent variants selected by mJAM 21 were novel (as defined in Methods ). Of the novel variants, 57 replicated at a Bonferroni level (p<0.05/184=0.00027, same direction of effect on PSA), an additional 80 replicated at p<0.05 (and the same direction), 43 demonstrated the same effect direction (but p>0.05), and four showed no indication of replication (effect in the opposite direction).…”

“…PRS-CSx relies on LD reference panels for estimating joint SNP effect sizes, while fine-mapping requires LD information for identifying independent variants from summary statistics. mJAM advances other fine-mapping approaches by incorporating population-specific LD, which is more accurate than using a single population as the LD reference 21 or making use of only the largest ancestry group. While PRS-CSx provides more flexibility to accommodate different genetic architectures, it may be more sensitive to the choice of LD reference panels and mismatches in LD structure between PRS training and testing populations, especially without a separate dataset for parameter tuning.…”

“…To identify independently associated variants (genome-wide significant, p≤5e-8) with computational efficiency, we first formed clumps of genome-wide significant variants such that all clumps were ≥10Mb apart and independent of one another; specifically, the top variant was chosen, genome-wide significant variants ≤10Mb from any variant in the clump were added to the clump, the process was iterated until a final clump was formed, and then the process was repeated to form more clumps (i.e., clumps were created such that there was no additional genome-wide significant variant ≤10Mb). Within each clump, we used mJAM v2022-08-05 21 , which uses population-specific linkage disequilibrium (LD) reference panels for each contributing cohort and ancestry group to model the correlation among variants, with an r 2 <0.01 threshold in all ancestry groups. Genotypes utilizing the appropriate GERA cohort group (European, Hispanic/Latino, African American, and East Asian) served as references 64 .…”

Improved multi-ancestry fine-mapping identifiescis-regulatory variants underlying molecular traits and disease risk

“…While existing multi-ancestry fine-mapping frameworks have been proposed for the analysis of complex traits and diseases 30,[33][34][35][36][37][38][39][40][41] , they have several limitations in the context of large-scale cis-molQTL data. First, many approaches do not model the correlation of causal variant effect sizes across ancestries or assume that they are a-priori independent across ancestries, which fails to reflect shared or similar genetic architectures 33,35,37,38 . Second, existing multi-ancestry approaches scale poorly, which precludes their application to thousands of molecular traits 3 commonly measured in cis-molQTL studies 33,35,40 .…”

“…Second, existing multi-ancestry approaches scale poorly, which precludes their application to thousands of molecular traits 3 commonly measured in cis-molQTL studies 33,35,40 . Third, current fine-mapping approaches lack ancestry-specific effect size estimates 33,35,37 , which neglects their potential use in post-Genome-wide Association Studies (GWASs) frameworks (e.g., Transcriptome-and Proteome-wide Association Studies (TWASs/PWASs) [42][43][44][45][46][47] . Last, while recent approaches address some of these limitations, existing software implementations are capable of analyzing only two ancestries, which excludes datasets consisting of ever-increasing diverse ancestries 39 .…”