A powerful microbiome-based association test and a microbial taxa discovery framework for comprehensive association mapping

Abstract

BackgroundThe role of the microbiota in human health and disease has been increasingly studied, gathering momentum through the use of high-throughput technologies. Further identification of the roles of specific microbes is necessary to better understand the mechanisms involved in diseases related to microbiome perturbations. MethodsHere, we introduce a new microbiome-based group association testing method, optimal microbiome-based association test (OMiAT). OMiAT is a data-driven testing method which takes an optimal test throughout different tests from the sum of powered score tests (SPU) and microbiome regression-based kernel association test (MiRKAT). We illustrate that OMiAT efficiently discovers significant association signals arising from varying microbial abundances and different relative contributions from microbial abundance and phylogenetic information. We also propose a way to apply it to fine-mapping of diverse upper-level taxa at different taxonomic ranks (e.g., phylum, class, order, family, and genus), as well as the entire microbial community, within a newly introduced microbial taxa discovery framework, microbiome comprehensive association mapping (MiCAM).ResultsOur extensive simulations demonstrate that OMiAT is highly robust and powerful compared with other existing methods, while correctly controlling type I error rates. Our real data analyses also confirm that MiCAM is especially efficient for the assessment of upper-level taxa by integrating OMiAT as a group analytic method.ConclusionsOMiAT is attractive in practice due to the high complexity of microbiome data and the unknown true nature of the state. MiCAM also provides a hierarchical association map for numerous microbial taxa and can also be used as a guideline for further investigation on the roles of discovered taxa in human health and disease.Electronic supplementary materialThe online version of this article (doi:10.1186/s40168-017-0262-x) contains supplementary material, which is available to authorized users.

A small‐sample kernel association test for correlated data with application to microbiome association studies

“…In this paper, we propose CSKAT to test the association between microbiome compositions and an outcome of interest, where microbiome samples and outcome measurements within the same cluster are related to each other. Similar to other methods (Koh et al, 2017;Wu, Chen, Kim, & Pan, 2016), our CSKAT framework can also be used as a taxon association mapping tool by shifting the analysis unit to a lower taxonomic rank (e.g., genus or family). Through extensive numerical studies, we have seen that CSKAT can protect the correct Type I error and achieve higher power than these existing methods, especially when the sample size is small or moderate.…”

“…Examples of related outcomes in microbiome studies include those from longitudinal and pedigree (family) studies (Borewicz et al, 2013;Goodrich et al, 2014;Lax et al, 2014;Morris et al, 2016;Turnbaugh et al, 2009). These methods usually assume unrelated samples and can lead to incorrect inference when there exists correlation among the outcomes.…”

“…sequencing technologies that enable measuring microbiome composition by direct DNA sequencing without the need for laborious cultivation (Lasken, 2012;Turnbaugh et al, 2007). Through analyses of large-scale microbiome studies, it has been found that the human microbiome contributes fundamentally to many human conditions, such as obesity (Turnbaugh et al, 2009), inflammatory bowel diseases (Morgan et al, 2015), colorectal cancer (Wang et al, 2012), Type II diabetes (Qin et al, 2012), and menopause-related symptoms (Mitchell et al, 2017).…”

A review of kernel methods for genetic association studies

“…In such situations, positive‐definiteness correction may be needed to improve the behavior of the kernel‐based score test. The distance‐based kernel design has been widely used for microbiome association studies such as MiRKAT (J. Chen & Li, ; Zhao et al, ), MMiRKAT (Zhan, Tong, et al, ), and OMiAT (Koh, Blaser, & Li, ).…”

“…The distance-based kernel design has been widely used for microbiome association studies such as MiRKAT (J. Chen & Li, 2013;Zhao et al, 2015), MMiRKAT (Zhan, Tong, et al, 2017), and OMiAT (Koh, Blaser, & Li, 2017).…”