Modeling intra-host dynamics within the microbiota
Résumé
Microbiota is a major reservoir of bacteria in the humans and animals organism. It is home to numerous commensal species, some of which can be sources of infection, such as staphylococcus aureus[1]. While the omposition and properties of microbiota are increasingly understood, their dynamics remain difficult to model, due to the large number of species and their interactions. Generalized Lotka Volterra (LV) model is particularly interesting, since it allows simulations of a large number of interacting microbial populations. However, calibrating this model requires abundance data, while classical metagenomic analyses, which quantify the composition of microbiota, only provide "frequency" data, i.e. the proportion of each population among those present. Currently, to address this issue, either imprecise proxies of total microbiota abundance are used [2], or strong assumptions are made about the system, for example by assuming that total abundance is fixed [3]. Applying this model to microbiota data without using such assumptions is therefore a key challenge:
- We characterize analytically the identifiability conditions of the LV model on frequency data.
- We demonstrate analytically that such identifiability is possible in the general case, without requiring strong assumptions.
- We validate this result by numerical analysis of simulations of microbiotic dynamics.
Origine : Fichiers produits par l'(les) auteur(s)
