Zero-inflated negative binomial mixed model: an application to two microbial organisms important in oesophagitis

Area of Science:

• Microbiology
• Bioinformatics
• Statistical Modeling

Background:

• Eosinophilic esophagitis (allergic inflammatory condition) is linked to changes in gut microbiota.
• High-throughput sequencing enables comprehensive identification of microbial communities.
• Microbiota data often exhibit complex characteristics like high zero counts and hierarchical structures.

Purpose of the Study:

• To present a hierarchical regression-based approach for analyzing microbiota sequence data.
• To evaluate factors associated with disease state in eosinophilic esophagitis.
• To address the challenges of zero-inflated and hierarchical data in microbiota studies.

Main Methods:

• Application of a zero-inflated negative binomial mixed model with random effects.
• Analysis of sequence count data from human microbiota studies.
• Evaluation of associations between specific organisms and disease state, adjusting for confounders.

Main Results:

• The study demonstrates a viable statistical approach for analyzing complex microbiota data.
• The hierarchical model effectively accounts for within-subject variation.
• Associations between specific microbial organisms and eosinophilic esophagitis were investigated.

Conclusions:

• Hierarchical regression models, specifically zero-inflated negative binomial mixed models, are suitable for analyzing microbiota data in diseases like eosinophilic esophagitis.
• This approach allows for robust evaluation of microbial associations while managing data complexities.
• Further application of such models can advance understanding of the role of microbiota in inflammatory conditions.