Fine‐Mapping Additive and Dominant SNP Effects Using Group‐LASSO and Fractional Resample Model Averaging

Abstract

Genomewide association studies sometimes identify loci at which both the number and identities of the underlying causal variants are ambiguous. In such cases, statistical methods that model effects of multiple SNPs simultaneously can help disentangle the observed patterns of association and provide information about how those SNPs could be prioritized for follow-up studies. Current multi-SNP methods, however, tend to assume that SNP effects are well captured by additive genetics; yet when genetic dominance is present, this assumption translates to reduced power and faulty prioritizations. We describe a statistical procedure for prioritizing SNPs at GWAS loci that efficiently models both additive and dominance effects. Our method, LLARRMA-dawg, combines a group LASSO procedure for sparse modeling of multiple SNP effects with a resampling procedure based on fractional observation weights; it estimates for each SNP the robustness of association with the phenotype both to sampling variation and to competing explanations from other SNPs. In producing a SNP prioritization that best identifies underlying true signals, we show that: our method easily outperforms a single marker analysis; when additive-only signals are present, our joint model for additive and dominance is equivalent to or only slightly less powerful than modeling additive-only effects; and, when dominance signals are present, even in combination with substantial additive effects, our joint model is unequivocally more powerful than a model assuming additivity. We also describe how performance can be improved through calibrated randomized penalization, and discuss how dominance in ungenotyped SNPs can be incorporated through either heterozygote dosage or multiple imputation.

A Permutation Approach for Selecting the Penalty Parameter in Penalized Model Selection

“…By contrast, when estimating , more permutations reduce variability without changing meaning. In our experience with simulated and real data, permutations is sufficient to control variability and provide reasonable values of ; in some cases, a lower N can suffice ( e.g ., in Valdar et al () and Sabourin et al (), the median of permutations was enough to select a stable model in the setting of fine‐mapping a GWAS locus).…”

ComPaSS‐GWAS: A method to reduce type I error in genome‐wide association studies when replication data are not available

“…One possible solution to the issue of losing a rare SNP on a split due to resampling may be to use an approach similar to the fractional resampling scheme used in Sabourin, Nobel, and Valdar (2015). A rare SNP can by chance have all subjects with the minor allele placed into the same half of the split thus ensuring it cannot be corroborated.…”

“…Although the loss of the rare SNP in a sample split is a plausible reason that it may not be replicated, this is not a desirable aspect for an analysis procedure. One possible solution to the issue of losing a rare SNP on a split due to resampling may be to use an approach similar to the fractional resampling scheme used in Sabourin, Nobel, and Valdar (2015). With this type of approach, rather than splitting the samples, each individual would have a regression weight ( U w~(0,1) i ) determining the contribution for analysis on D A and have a complementary weight (1 − w i ) for the analysis on D B .…”

“…Such a modification would ensure each SNP is included on each analysis, but the conditional independence between D A and D B on each split is lost. Sabourin et al (2015) demonstrated that in a resampling procedure based on just D A , fractional resampling performed nearly identical to subsampling half the data, but further testing would be necessary to see how it would affect the complementary pairs based analysis. SABOURIN ET AL.…”

An approximate Bayesian significance test for genomic evaluations

“…Bennewitz et al, 2017) and for other SNP-selection approaches (e.g. Sabourin, Nobel, & Valdar, 2015). However, sample size matters when nonadditive effects are studied (e.g.…”