Keywords: cheese ecosystem, mathematical modeling, differential analysis, RNA-seq, resilience, deterministic model.

The emergence of "omics" methods is an opportunity to understanding cheese microbial ecosystems functions, and their resilience, which represents a major scientific and economic challenge. To study its resilience, capacity for an ecosystem to maintain its function following a perturbation, a reduced and controlled ecosystem of a washed rind cheese [1] was challenged by modifying either the salinity of the curd or by omitting one of the two major yeast species. Metatranscriptomics and physiochemical data was acquired during each ripening. We propose a two-step approach combining these different kind of data to study the resilience and apply it on casein proteolysis.

In order to identify the role of each species in the proteolysis resilience, we have developed a kinetic model integrating the biochemical data of proteolysis and the population dynamics. Model parameters were estimated by Metropolis-Hastings algorithm and MCMC chains using the MCMC Matlab Toolbox [2]. With this approach, we have highlighted the major role of the yeasts Debaryomyces hansenii and Geotrichum candidum in this phenomenon.

Then we have focused transcriptomic analysis on these major species. We selected genes declared differentially expressed using the R package DESeq2 [3] between a disturbed condition and the normal one at 5% after correcting for multiple testing. As the number of selected genes was exceeding the number of observations, we have used a lasso penalized regression model to identify genes potentially predictors for the proteolysis.

Our approach has identified a set of implicated genes. This work is an example of an approach which integrates diverse data (microbial growth, biochemical, physiochemical and genomic data) in order to understand the microbial ecosystem resilience. Our approach is a general one for studying microbial ecosystems with high-throughput methods and has shown the feasibility of a functional microbial ecology.

[1] Mounier, C. Monnet, T. Vallaeys, R. Arditi, A.S. Sarthou, A. H elias, and F. Irlinger. Microbial interactions within

a cheese microbial community. Applied and environmental microbiology, 74(1) :172–181, 2008

[2] H. Haario, M. Laine, A. Mira and E. Saksman, 2006. DRAM: Efficient adaptive MCMC, Statistics and Computing 16, pp.

339-354.

[3] Simon Anders, Wolfgang Huber: Differential expression analysis for sequence count data Genome Biology 11 (2010) R106, J.