MCSPACE: inferring microbiome spatiotemporal dynamics from high-throughput co-localization data.

Area of Science:

• Microbiome research
• Computational biology
• Systems biology

Background:

• High-throughput sequencing methods like SAMPL-seq enable large-scale characterization of gut microbiome biogeography.
• Analyzing high-dimensional microbiome co-localization data presents significant computational challenges due to data complexity and noise.

Purpose of the Study:

• To develop a probabilistic AI method, MCSPACE, for inferring microbial assemblages and their dynamics from co-localization data.
• To address the complexity and noise inherent in high-dimensional microbiome data.

Main Methods:

• Development of MCSPACE, a probabilistic AI tool for analyzing microbiome co-localization data.
• Generation of a large longitudinal mouse gut microbiome co-localization dataset with dietary perturbations.
• Validation using existing human longitudinal datasets.

Main Results:

• MCSPACE successfully infers spatially coherent microbial assemblages, their temporal dynamics, and responses to perturbations.
• Benchmarking demonstrated MCSPACE's superior performance compared to existing methods.
• Identification of persistent and dynamic microbial assemblages in the human gut and diet-induced shifts in murine gut assemblages.

Conclusions:

• MCSPACE is a valuable open-source tool for studying microbiome biogeography dynamics.
• The method provides insights into the role of spatial relationships in host-microbial ecosystem function.
• Elucidating microbiome spatial structuring is crucial for understanding host-microbial interactions.