mRNA biomarkers selection based on Partial Least Square algorithm in order to further predict Bacillus weihenstephanensis acid resistance

In order to integrate omics data to quantitative microbiological risk assessment in foods, gene expressions may serve as bacterial behaviour biomarkers. In this study an integrative approach encompassing predictive modelling and mRNAs quantifications, was followed to select molecular biomarkers to further predict the acid resistance of Bacillus weihenstephanensis. A multivariate analysis was performed to correlate the acid bacterial resistance and the gene expression of vegetative cells with or without exposure to stressing conditions. This mathematical method provides the advantage to take gene expressions and their interactions into account. The use of the Partial Least Squares algorithm allowed the selection of nine genes as acid resistance biomarkers among thirty targeted genes. According to their involvement in the general acid stress response of Bacillus, these genes were assigned to three different biological modules namely, metabolic rearrangements, general stress response and oxidative stress response. The oxidative stress response appeared as the major activated biological module in B. weihenstephanensis cells submitted to acid stress conditions. Furthermore, as a firstly described model, the developed concept showed promising results to further be used to predict bacterial resistance using gene expression. Thus, this study underlines the possibility to integrate the bacterial physiology state, using omics biomarkers, into bacterial behaviour modelling and provide mechanistic understanding in acid bacterial resistance mechanisms.