Related Experiment Video
Updated: Jan 1, 2026
07:21
Tick Microbiome Characterization by Next-Generation 16S rRNA Amplicon Sequencing
Published on: August 25, 2018
13.3K
A Bayesian zero-inflated negative binomial regression model for the integrative analysis of microbiome data.
Shuang Jiang1, Guanghua Xiao2, Andrew Y Koh3
1Department of Statistical Science, Southern Methodist University, Dallas, TX 75275, USA.
Biostatistics (Oxford, England)
|December 18, 2019
Summary
This study introduces a new Bayesian regression model to analyze the human microbiome. The model integrates microbiome data with other factors to identify disease-associated bacteria and understand their interactions.
Area of Science:
- Microbiology
- Bioinformatics
- Statistical Modeling
Background:
- Microbiome omics studies reveal links between microbial communities and diseases.
- Metabolism, genetics, and environment influence microbiome effects.
- Current methods for integrating microbiome data with covariates are limited.
Purpose of the Study:
- To develop an integrative Bayesian regression model for microbiome analysis.
- To distinguish differentially abundant taxa associated with phenotypes.
- To quantify the effects of covariates on microbiome taxa.
Main Methods:
- Developed an integrative Bayesian zero-inflated negative binomial regression model.
- Applied the model to simulated data to assess performance.
- Integrated microbiome taxonomies and metabolomics in real-world datasets.
Main Results:
- The proposed model demonstrated good performance with simulated data.
- Successfully integrated microbiome and metabolomics data from two real datasets.
- Generated biologically interpretable findings from integrated analyses.
Conclusions:
- A novel integrative Bayesian regression model for microbiome studies is proposed.
- The model enables bacterial differential abundance analysis.
- The model quantifies microbiome-covariate effects, suitable for general microbiome research.
