Comparison of statistical tests for disease association with rare variants

Abstract

In anticipation of the availability of next-generation sequencing data, there is increasing interest in investigating association between complex traits and rare variants (RVs). In contrast to association studies for common variants (CVs), due to the low frequencies of RVs, common wisdom suggests that existing statistical tests for CVs might not work, motivating the recent development of several new tests for analyzing RVs, most of which are based on the idea of pooling/collapsing RVs. However, there is a lack of evaluations of, and thus guidance on the use of, existing tests. Here we provide a comprehensive comparison of various statistical tests using simulated data. We consider both independent and correlated rare mutations, and representative tests for both CVs and RVs. As expected, if there are no or few non-causal (i.e. neutral or non-associated) RVs in a locus of interest while the effects of causal RVs on the trait are all (or mostly) in the same direction (i.e. either protective or deleterious, but not both), then the simple pooled association tests (without selecting RVs and their association directions) and a new test called kernel-based adaptive clustering (KBAC) perform similarly and are most powerful; KBAC is more robust than simple pooled association tests in the presence of non-causal RVs; however, as the number of non-causal CVs increases and/or in the presence of opposite association directions, the winners are two methods originally proposed for CVs and a new test called C-alpha test proposed for RVs, each of which can be regarded as testing on a variance component in a random-effects model. Interestingly, several methods based on sequential model selection (i.e. selecting causal RVs and their association directions), including two new methods proposed here, perform robustly and often have statistical power between those of the above two classes.

Phenotypic extremes in rare variant study designs

“…8,9 The power to detect association with the various proposed gene-based methods is dependent on the underlying genetic architecture of the gene. [10][11][12] The relative power of different study designs for CVASs has been well established. 13 For all genetic studies, selecting the extremes of the phenotype distribution improves power; a concept in genetics that can be traced back to seminal work by Lander and Botstein.…”

A Powerful and Adaptive Association Test for Rare Variants

“…Hence, developing new association tests tailored to RVs has been an active research area in the past few years. Due to low MAFs of RVs, to achieve practically meaningful power, the majority of existing approaches focus on testing on a group of RVs, rather than on each individual RV (Capanu et al 2011); the main idea is to boost power through aggregating information across multiple RVs in an analysis unit, such as a gene (e.g., Morgenthaler and Thilly 2007;Li and Leal 2008;Madsen and Browning 2009;Liu and Leal 2010;Han and Pan 2010;Hoffmann et al 2010;Li et al 2010;Price et al 2010;Zhang et al 2010;Zhu et al 2010;Luo et al 2011;Neale et al 2011;Ionita-Laza et al 2011;Feng et al 2011;Basu and Pan 2011;Gordon et al 2011;Wu et al 2011;Fan et al 2013). As theoretically shown (Cox and Hinkley 1974) and demonstrated in our simulations, there is no uniformly most-power test for this purpose, which means that, depending on the unknown truth, including specific association effect directions and sizes, a given and fixed test may or may not be powerful.…”

A Biologically Informed Method for Detecting Associations with Rare Variants

“…The CMC method uses a multivariate statistical test and permits combined analysis of rare and common variants [21]. The CMC method has improved power over CAST, presumably because functional information (direction of effect) was incorporated and because the method can be implemented in a regression framework [54].…”

“…This is currently performed using filtering methods, such as removing variants found in publicly available databases collected from "healthy" individuals or binning only nonsynonymous or predicted damaging variants. Other methods use "step-up" approaches or "sliding windows" to intermediately test the strength of the association signal, and prune out variants that do not positively impact the strength of association [49,54].…”