Modeling Fiber Degradation by Human Gut Microbiome
Résumé
Description of the human microbiome is a critical issue. Standard approaches in gut microbial ecology are based on characterization of the taxonomic range and diversity of the main species sampled from feces. Paradoxically, while a very large taxonomic diversity has been observed between individuals, gut carries the same metabolic functions accross humanity. This can be motivated by ecological arguments: bacteria undergo a strong environmental selection pressure in the human gut which results in a low functional diversity. In this context, describing functional traits in microbiomes of a variety of individuals appears as a powerful approach to describe microorganisms selected in the gut. Thus, we aim at characterizing the main fiber degradation traits using recently available metagenomic data and tools adapted from blind source separation. 1-Reaction Graph and Metagenomic Data Reaction Graph: A strenuous work of selection is carried out in order to identify metabolites and reactions which are specifically involved in anaerobic catabolism pathways, as well as the known associated genes. We obtain 25 GH and PL families and an acyclic graph of 43 metabolites and 61 reactions representing sugar degradation (Figure 1). This graph is characterized by its incidence matrix Q ∈ R 43×61 , such that Q ij = 1 if edge j reaches vertex i, Q ij = −1 if edge j leaves vertex i and Q ij = 0 otherwise. Q is split into its positive and negative parts as Q = Q + − Q -