What question did this study set out to answer?

To present BCGLMs, a package designed for Bayesian modeling of disease predictions using microbiome data.

February 13, 2026Open Access

BCGLMs: Bayesian Modeling for Disease Prediction Using Compositional Microbiome Features

Key Points

To present BCGLMs, a package designed for Bayesian modeling of disease predictions using microbiome data.
Developed an R package for Bayesian compositional modeling.
Incorporated various response types including continuous, binary, ordinal, and survival.
Implemented random effects to enhance prediction accuracy.
Used the brms package as a foundation to facilitate model fitting.
Demonstrated improved prediction accuracy with random effects in models.
Provided tools for both numeric and graphical summarization of model results.

Abstract

Abstract Motivation BCGLMs is a freely available R package that provides functions for setting up and fitting Bayesian compositional models for continuous, binary, ordinal and survival responses. It also includes models with random effects to capture sample-related accumulated small effects, improving prediction accuracy. The package includes tools for summarizing results from fitted models both numerically and graphically. Built on top of the widely used brms package, BCGLMs enable users to incorporate phylogenetic relationships between microbiome taxa into the modeling framework. Overall, BCGLMs package offers a flexible and powerful set of tools for analyzing compositional microbiome data. Availability and implementation The package is publicly available via GitHub https://github.com/Li-Zhang28/BCGLMs.

BCGLMs: Bayesian Modeling for Disease Prediction Using Compositional Microbiome Features

Key Points

Abstract

Cite This Study