Dynamic Bayesian Networks for Integrating Multi-omics Time Series Microbiome Data.

Area of Science:

• Microbiome research
• Systems biology
• Computational biology

Background:

• Longitudinal microbiome data analysis is challenging.
• Inferring temporal interactions between microbial taxa, genes, metabolites, and host genes is complex.
• Existing methods for joint modeling of multi-omics data are limited.

Purpose of the Study:

• To develop a computational pipeline for analyzing longitudinal multi-omics data.
• To integrate diverse data types including microbiome sequence, gene expression, and metabolomics.
• To reconstruct a unified model of temporal interactions.

Main Methods:

• Developed a pipeline for the analysis of longitudinal multi-omics data (PALM).
• Utilized dynamic Bayesian networks (DBNs) to reconstruct a unified model.
• Aligned multi-omics data, overcoming differences in sampling and progression rates.

Main Results:

• PALM accurately identifies known and novel interactions in inflammatory bowel disease patient data.
• The pipeline successfully integrated sequence, expression, and metabolomics data.
• Experimental validations supported predicted novel metabolite-taxon interactions.

Conclusions:

• PALM is an effective tool for joint modeling of longitudinal multi-omics microbiome data.
• The pipeline can identify complex interactions impacting host gene expression.
• PALM facilitates the discovery of novel biological relationships in microbiome studies.