Biomass distances from personalized multispecies dynamic flux balance analysis of the human gut microbiome identify dietary influences for patients with and without inflammatory bowel disease

A parallelized version of a multispecies dynamic flux balance analysis (msdFBA) algorithm is implemented and applied to the AGORA collection of genome-scale metabolic reconstructions for 818 members of the human gut microbiome. The msdFBA method assumes the well stirred interaction mode of all organisms to exchange external metabolites. In each msdFBA simulation, the biomasses the gut microbiome composition of one of 149 patients from NIH Human Microbiome Project is used for initialization in combination with one of 11 different diets used as substrates as defined in the Virtual Metabolic Human database. The union of all species in the patient data comprises 255 different microbes. The patients are either healthy or suffer from inflammatory bowel disease (IBD). The msdFBA simulation is performed for 50 time steps. For all combinations of patients and time steps, the euclidean distance between the vector of the biomasses of the 255 patient species and the evolving vector of biomasses for the same species is calculated, providing the information about the biomass distance to each patient during each simulation. To quantify the overall influence of a diet for all patients, a diet score is defined as the sum of the reciprocal distances to the closest patient at the last time step, in case the closest patient is diseased, subtracted from the respective sum for the case that the closest patient is healthy. With this score, the known beneficial influences both of a high fiber and a gluten free diet for IBD is verified. Noteworthy is the utility of a Mediterranean diet in this context, having similar distance patterns. The proposed method provides an universal platform for the in-silico analysis of different environmental influences like diets for different microbiotas defined by metagenomic quantifications from individual patients and has the potential to generate additional dietary recommendations for the management of various other diseases.