Two dynamic regimes in the human gut microbiome.

Area of Science:

• Microbiology
• Systems Biology
• Computational Biology

Background:

• The gut microbiome is a complex, dynamic ecosystem influenced by host factors and microbe interactions.
• Previous studies on gut microbial dynamics primarily used autoregressive models, like Lotka-Volterra.
• Community state-clustering methods may oversimplify within-host microbial dynamics.

Purpose of the Study:

• To investigate the non-autoregressive components of gut microbial time series.
• To evaluate the limitations of clustering methods for within-host dynamics.
• To model gut microbial dynamics using a novel approach suitable for non-autoregressive variance.

Main Methods:

• Application of sparse vector autoregression (sVAR), an econometrics-derived method, to model microbial time series.
• Analysis of microbial variance to distinguish autoregressive and non-autoregressive components.
• Correlation analysis between microbial taxa, diet, and phylogenetic signals.

Main Results:

• Most gut microbial variance is non-autoregressive, challenging traditional models.
• Dietary changes exhibit non-autoregressive dynamics and significantly impact microbial variance.
• Phylogenetic signals in residuals suggest niche filtering as a key ecological process.
• Internal processes, possibly related to redox fluctuations, drive autoregressive variance.

Conclusions:

• The human gut microbiota exhibits two distinct dynamic regimes: one influenced by external factors like diet, and another by internal microbial processes.
• Non-autoregressive modeling is crucial for accurately characterizing gut microbial dynamics.
• Diet is a primary driver of gut microbial fluctuations, while internal mechanisms regulate stability.