Title: Variance component selection with microbiome taxonomic data
Authors: Jin Zhou - University of Arizona (United States) [presenting]
Jing Zhai - University of Arizona (United States)
Hua Zhou - UCLA (United States)
Abstract: High-throughput sequencing technology has enabled population-based studies of the role of the human microbiome in disease etiology and exposure response. One important problem in microbiome analysis is to identify the bacterial taxa that are associated with a response, where the microbiome data are summarized as the counts or composition of the bacterial taxa at different taxonomic levels. Previous methods consider variable selection of taxonomic data in regression analyses and taxonomic data at different levels are considered as fixed covariates. Due to high-dimensional features of metagenomic information, different penalization schemes have been adopted. On the other hand, the association of microbiome composition and clinical phenotypes was assessed by testing the nullity of variance components, where phylogenetic tree information and distances measures between communities can be incorporated into the model. By combining these two methods we consider regression analysis by treating bacterial taxa at different level as multiple random effects. We propose a variance component selection scheme of high-dimensional taxonomic clusters with Lasso penalization. Our methods seamlessly coupled different distance measures with associated taxonomic selection. Extensive simulations demonstrate the superiority of our methods vs existing methods. Finally, we apply our method to a longitudinal lung microbiome study of HIV infected patients.