Academic journal article Genetics

A Powerful and Adaptive Association Test for Rare Variants

Academic journal article Genetics

A Powerful and Adaptive Association Test for Rare Variants

Article excerpt

(ProQuest: ... denotes formulae omitted.)

THE recent advances in sequencing technologies have made it feasible to conduct global testing for association between complex traits and rare variants (RVs) (Bansal et al. 2010). The most popular approach in genome-wide associ- ation studies (GWASs) is to test on each single nucleotide variant (SNV) one by one and then select the SNVs meeting a stringent signi ficance level after adjusting for multiple testing. However, such a strategy may be low powered due to the weak signal contained within each individual RV for its extremely low minor allele frequency (MAF). Hence, develop- ing 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 RVsinananalysisunit,suchasagene(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; Pan and Shen 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 demon- strated 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. Hence, there have been intensive efforts in developing adap- tivetestsforRVs(e.g., Pan and Shen 2011; Lin and Tang 2011; Zhang et al. 2011; Lee et al. 2012; Chen et al. 2012; Derkach et al. 2013; Sun et al. 2013). However, due to their limited extents of adaptivity (e.g., with a predetermined and fixed set of the weights on RVs), these adaptive tests are still not flexible (or adaptive) enough with loss of power in some situations. A main motivation in this article is to develop a broader family of association tests such that at least one of them is powerful for a given situation. We develop such a fam- ilyoftests,calledthesumofpoweredscore(SPU)tests,which generalize the sum (of score) test (Sum) and the sum of squared score (SSU) test (Pan 2009). The Sum test is a repre- sentative of the burden tests based on genotype pooling or collapsing (Morgenthaler and Thilly 2007; Li and Leal 2008; Madsen and Browning 2009), whereas the SSU test is closely related to kernel machine regression [and its implementation for RVs, SKAT (Sequence Kernel Association Test)] (Wu et al. 2010, 2011), C-alpha test (Neale et al. 2011), and an empirical Bayes test for high-dimensional data (Goeman et al. 2006); see Pan 2011 and Basu and Pan (2011). In many simulation set- ups, one, but not both, of the Sum test and SSU test has been shown to be powerful (Basu and Pan 2011). For example, with different association directions of causal RVs, the Sum test suffers from a loss of power, while the SSU test performs much better. However, we emphasize that, in analysis of multiple RVs, there exist nonassociated RVs. For example, in cancer research it has been observed that the vast majority of RVs do not appear to confer risk (Capanu et al. 2011). Hence, it is important to assess the performance of a test in the presence of nonassociated RVs in the group of the RVs to be tested. In fact, as to be shown, the performance of the Sum test deteriorates rapidly as the number of nonassociated RVs increases, whereas the SSU test is more robust but nevertheless may gradually become less competitive. It seems that the performance of various tests has not been fully investigated for the case with many nonassociated RVs, including some new adaptive tests, such as a kernel-based adaptive clustering (KBAC) test (Liu and Leal 2010), a P-value weighted sum test (PWST) (Zhang et al. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.