The human intestinal microbiota is a complex microbial ecosystem that plays a crucial role in several aspects of human health. It is particularly involved in the metabolism of residual fibers, through anaerobic digestion, thus providing significant energy (Short Chain Fatty Acids, simple sugars) and vitamins to the host. Whole Genome Sequencing (WGS) data from metagenomic analyses give an insight of the content in terms of genes of an entire microbial community, even if the organisms that compose it cannot be cultivated. In addition to the potential of conventional molecular inventory techniques (such as targeting ADNr16s), which allows an analysis of diversity, WGS approaches provide an access to the functions.

Our main goal is to integrate this functional information in a previously developed higher level mechanistic model of the microbiota carbohydrate trophic chain[1]. Thus, we investigate how the concept of functional population is relevant in order to synthesize WGS based information.

We formulate a problem similar to a regularized Blind Source Separation (BSS) approach which allows us to create functional profiles. As we want theses profiles to be biologically meaningful, the regularization term is critical since it let us incorporate prior information during the profiles creation process. Therefore we explore a variety of classic penalization (sparsity, lasso and elastic net lasso) while developing our original criteria.

[1] R. Muñoz-Tamayo, B. Laroche, Éric Walter, J. Doré, and M. Leclerc. Mathematical modelling of carbohydrate degradation by human colonic microbiota. J. Theor. Biol., 266(1): 189-201, 2010.