Recent advances in DNA sequencing and metagenomics are opening a window into the human microbiome, and revealing novel associations between microbes, health and disease. But most microbiome studies are cross-sectional and lack a mechanistic understanding of this ecosystem.

We developed a method to analyze dynamics of microbiome composition which accounts for time-dependent external perturbations such as antibiotics. The new method combines Lotka–Volterra models of population dynamics with regression techniques and can be used to predict ecosystem dynamics.

We demonstrate the model using data from mouse experiments and we show that we can recover the microbiota temporal dynamics and study the concept of alternative stable states and antibiotic-induced transitions. The model suggests that a small group of commensal microbes protects against infection by the pathogen Clostridium difficile and explains how the antibiotic makes the host more susceptible to infection by perturbing the protective consortium.

Ecological modeling from time-series inference: insight into dynamics and stability of intestinal microbiota
Richard R. Stein*, Vanni Bucci*, Nora C. Toussaint, Charlie G. Buffie, Gunnar Rätsch, Eric G. Pamer, Chris Sander, João B. Xavier. PLoS Computational Biology
[Article: open access]