Estimating the Power of Detecting Microbiome Associations in Epidemiological Studies by Incorporating Temporal Stability

• Activity Number: 453
• Type: Contributed
• Date/Time: Wednesday, August 6, 2014 : 8:30 AM to 10:20 AM
• Sponsor: ENAR
• Author(s): Jianxin Shi*+
• Keywords: Microbiomics ; temporal stability ; power estimate

Abstract

The human microbiota is the collection of microbes inhabiting the human body. Technical advances have revolutionized the characterization of human microbiota and made it possible to perform large-scale epidemiological studies to associate the microbiota with complex human diseases. One important question for designing an epidemiological study is to evaluate the statistical power for detecting associations. Recent studies have shown substantial temporal variability for microbial communities in different body sites, suggesting that methods not incorporating such temporal instability may severely overestimate the power. We develop a data generative model for analyzing microbiome sequence data with multiple sampling points, which provides a natural statistical framework to evaluate temporal stability and statistical power for detecting associations. We estimated the parameters for both common and rare taxa for human microbiota in 18 body sites using the data in the Human Microbiome Project. Base on the estimated parameters, our method provide a more realistic estimate of the statistical power in microbiome-based epidemiological studies.