Predicting the carbon source for Bacillus subtilis by integrating gene expression profiles into a constraint-based metabolic model

Finding drug targets for antimicrobial treatment is a central focus in biomedical research. To discover new drug targets, we are interested in finding out which nutrients are essential for pathogenic microorganisms in the host or under specific circumstances. Besides, metabolic fluxes have been successfully constructed and predicted by employing flux balance analysis (FBA) technique. While 13C metabolic data is the most informative way to explore metabolism, the data can be difficult to acquire in complicated environments, for example, osteomyelitis from S. aureus. On the other hand, although gene expression data is less informative in this case as compared to 13C metabolic data, it is easier to generate, and it still provides us informative insights. We develop FBA models using the stoichiometric knowledge of the metabolic reactions of a cell and combine them with gene expression profiles. We aim to identify essential drug targets for specific nutritional uptakes of pathogenic microorganisms. As a case study, we implemented our method by applying data from B. subtilis to predict carbon sources based on given gene expression profiles. We validated our flux prediction results by comparing with 13C metabolic flux data. With our method, we efficiently utilized gene expression profiles to predict carbon sources and investigate the metabolic network of B. subtilis. We show that our method is promising, generalizable, and versatile. We present that using FBA model with gene expression data is a good starting point to support subsequent hypotheses to conduct further studies; especially, in the environment that 13C metabolic flux data is hard to achieve. Besides, from a technical aspect, our method performed faster in order to remove thermodynamically infeasible loops as compared to loopless COBRA (ll-COBRA), which is the well-established method in the community.