Compositional Knockoff Filter for FDR Control in Microbiome Regression Analysis

• Activity Number: 461 - Bugs, Bugs Everywhere - the Statistics Behind Our Microbiome
• Type: Topic Contributed
• Date/Time: Wednesday, August 1, 2018 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistical Graphics
• Author(s): Arun Srinivasan* and Lingzhou Xue and Xiang Zhan
• Companies: Pennsylvania State Univ and Penn State University and National Institute of Statistical Sciences and Pennsylvania State University
• Keywords: Knockoff Filter; FDR; Lasso; Microbiome compositional data

Abstract

A critical task in microbiome analysis is to identify microbial taxa that are associated with a response of interest. Most existing statistical methods examine the association between the response and one bug at a time, then followed by multiple testing adjustments such as false discovery rate (FDR) control. Despite feasibility, these methods are often underpowered due to unique characteristics of microbiome data, such as high-dimensionality, compositional constraint, and complex correlation structure. In this paper, we adopt the Knockoff Filter to provide finite sample false discovery rate control in the context of linear log-contrast models for regression analysis of compositional data. Instead of applying multiple testing corrections to many individual p-values, our framework achieved the FDR control in a regression model that jointly analyzes the whole microbiome community. By imposing an L1 regularization in the regression model, a subset of bugs is selected as related to the response under a preset FDR threshold. The method is demonstrated via simulation studies and is illustrated by an application to a recent study relating microbiome composition to host gene expressions.