|
Activity Number:
|
244
- Statistical methods for microbiome data analysis and beyond
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 11, 2021 : 10:00 AM to 11:50 AM
|
|
Sponsor:
|
Section on Statistics in Genomics and Genetics
|
|
Abstract #318364
|
|
|
Title:
|
Identifying Bugs Over Time: A Method Comparison for Association Tests in Longitudinal Microbiome Data
|
|
Author(s):
|
Nicholas E Weaver* and Audrey E Hendricks and Brandie D Wagner
|
|
Companies:
|
University of Colorado Denver and University of Colorado Denver and Colorado School of Public Health-Biostatistics and Informatics
|
|
Keywords:
|
Microbiome;
Longitudinal Data Analysis;
Simulations;
Beta-Diversity;
Method Comparison;
RMarkdown
|
|
Abstract:
|
In the past two decades the scientific community has made great strides in studying the relationship between the microbiome and human health. Statistical tools have been developed and applied to better understand the role of microbial communities throughout the body (e.g. in the gut and lungs) and across health and disease (e.g. cancer and obesity). However, most method development and comparison for microbiome data has been performed with data collected at one point in time. With longitudinal microbiome data becoming more prevalent throughout biomedical research, there is a need to identify and develop optimal methods within this more complex study design. Here we compare the ability of PERMANOVA, CSKAT and two-stage generalized linear mixed models to detect longitudinal association between the microbiome and a continuous phenotype. We compare the power and type I error from simulations derived from real longitudinal microbiome data. We offer some recommendations for where each method is appropriate and optimal. A detailed RMarkdown vignette of our simulation procedure is provided as an exemplar to increase accessibility of simulating realistic longitudinal microbiome data.
|
Authors who are presenting talks have a * after their name.
