Genetic fine-mapping from summary data using a nonlocal prior improves the detection of multiple causal variants

Abstract

Motivation: Genome-wide association studies (GWAS) have been successful in identifying genomic loci associated with complex traits. Genetic fine-mapping aims to detect independent causal variants from the GWAS-identified loci, adjusting for linkage disequilibrium patterns.

Results: We present "FiniMOM" (fine-mapping using a product inverse-moment prior), a novel Bayesian fine-mapping method for summarized genetic associations. For causal effects, the method uses a nonlocal inverse-moment prior, which is a natural prior distribution to model non-null effects in finite samples. A beta-binomial prior is set for the number of causal variants, with a parameterization that can be used to control for potential misspecifications in the linkage disequilibrium reference. The results of simulations studies aimed to mimic a typical GWAS on circulating protein levels show improved credible set coverage and power of the proposed method over current state-of-the-art fine-mapping method SuSiE, especially in the case of multiple causal variants within a locus.

Availability and implementation: https://vkarhune.github.io/finimom/.

Figure 1.

Comparison of a marginal inverse-moment (iMOM, solid black line) distribution with $\tau =0.00385$ and a Gaussian (N, dashed red line) distribution with mean 0 and variance ${0.1}^2$.

Figure 2.

95% credible set coverage (upper panel) and power (lower panel) and their 95% confidence intervals in the Northern Finland Birth Cohort 1966 simulation study (calculated over 100 simulation replicates) using FiniMOM with different values for hyperparameters τ and u. τ controls the spread of the detectable effect sizes, and u controls the prior for model dimension. Larger values of τ correspond to larger causal effect sizes that can be detected, while larger values for u refer to stronger priors toward smaller dimensions. The parameter value combinations used in the subsequent comparisons with SuSiE are highlighted with a box. LD: linkage disequilibrium; SuSiE: sum-of-single-effects.

Figure 3.

Comparison of credible set coverage and their 95% confidence intervals for FiniMOM and SuSiE in the simulation study (calculated over 100 simulation replicates). The grey dashed line represents the nominal 95% target coverage. LD: linkage disequilibrium; FiniMOM: Fine-mapping using inverse-moment priors; SuSiE: sum of single effects.

Figure 4.

Comparison of credible set power and their 95% confidence intervals for FiniMOM and SuSiE in the simulation study (calculated over 100 simulation replicates). LD: linkage disequilibrium; FiniMOM: fine-mapping using inverse-moment priors; SuSiE: sum of single effects.

Figure 5.

Locus plot of genetic associations ($-{\hspace{0.17em}\text{log}\hspace{0.17em}}_{10}(p)$) per each variant within $\mathit{NLRC}4$ locus (±1 Mb from rs385076 variant), with credible sets highlighted for FiniMOM (left panels) and SuSiE (right panels) in the discovery GWAS (top panels) and in the replication GWAS (bottom panels). FiniMOM: fine-mapping using inverse-moment prior; SuSiE: sum of single effects; GWAS: genome-wide association study.