MODELING MICROBIAL ABUNDANCES AND DYSBIOSIS WITH BETA-BINOMIAL REGRESSION

Area of Science:

• Microbiome research
• Statistical modeling
• Bioinformatics

Background:

• Estimating taxon relative abundance in microbial populations is essential.
• Existing models do not account for covariate-associated overdispersion.

Purpose of the Study:

• Propose a novel beta-binomial model for microbiome analysis.
• Develop methods to test for differential relative abundance and differential variability.
• Investigate the link between dysbiosis and changes in microbial community stability.

Main Methods:

• Developed a beta-binomial model incorporating covariate-associated overdispersion.
• Proposed statistical tests for differential relative abundance and variability.
• Validated the model using simulation studies and soil microbial data.

Main Results:

• The proposed model effectively estimates taxon relative abundance.
• The model successfully identifies differential variability, a potential indicator of dysbiosis.
• Demonstrated the model's utility in analyzing soil microbial communities.

Conclusions:

• The beta-binomial model offers a powerful tool for microbiome analysis.
• Detecting differential variability provides new insights into microbiome stability and dysbiosis.
• This approach advances our understanding of microbial ecology and disease states.