Fig 1.
Workflow of visualization of metabolic networks and associated data.
This chart describes the flow of data inputs and analyses in VisANT 5.0, with focus on functionalities which help interpret community-level models of microbial metabolism. VisANT 5.0 integrates three main data types (top pink layer) pertaining to ecosystem-level metabolism: (i) One or more stoichiometric models of metabolic networks for different organisms; (ii) Flux data for all reactions in all organisms at different time points; (iii) A layout file where spatial information of metabolic models is specified. Adding these different types of information on top of each other leads to visualization of different features of the network (gray layer). These networks can be combined with additional knowledge on the catalyzing enzymes, such as regulatory effects or protein-protein interaction edges (bottom layer, green).
Fig 2.
VisANT visualization of metabolic cross-feeding between two bacteria, using the new “Symbiotic Layout” functionality.
This specific system is a previously evolved, obligate syntrophic consortium between a genetically modified E. coli strain which requires an external supply of methionine, and a S. enterica strain that cannot use the only carbon source available in the environment (lactose). The system was simulated with COMETS (Computation of Microbial Ecosystems in Time and Space), and represented in VisANT. For this case study we used a single spatial point (i.e. a 1 by 1 grid in COMETS), thus loading one grid point or the entire simulated COMETS grid are equivalent. Models are represented as expanded metanodes, exchange reactions are shown as nodes with (X,Y) graphs representing the flux through them throughout the simulated growth experiment, with an arrow denoting the current time step. Extracellular metabolite nodes are color coded based on the type of interaction they mediate: (i) Blue if it is secreted by both organisms; (ii) Red if it is consumed by both; (iii) Light gray if one model produces it and the other consumes it, and (iv) Dark gray if the metabolite is only associated with one model. E. coli can be seen here taking up lactose, and secreting acetate as a by-product. S. enterica, in turn, is able to grow using the acetate, and secretes methionine, which allows E. coli to continue to grow. Users can trace through the network by double-clicking a node to reveal connected nodes. This was used to trace lactose through the E. coli network, and to display some of the intracellular reactions adjacent to the exchange fluxes in S. enterica. In order to make the differences in fluxes more apparent for the reactions that mediate interactions between the two species, all the corresponding fluxes were mapped logarithmically onto the edge weights, using the "Rescale Selected Edges Logarithmically” function (see User Manual–Rescale Selected Edges Logarithmically at Page 10, S1 Text).
Fig 3.
Metabolic exchange in a microbial ecosystem.
Visualization of metabolic exchange between many organisms in a system is achieved through VisANT's multi-organism layout. The six organisms shown here are microbes selected because of their roles and abundance in the human gut. Metabolic flux was determined through a COMETS simulation involving all six microbes in a minimal D-glucose media. Each model is represented by a metanode, and in the center of the models are nutrients in the media that interact with the models. Nodes are color coded by their syntrophic influence, gray being nutrients involved in a potential syntrophy, red being nutrients which the microbes may be competing over, blue being nutrients which are produced by more than one organism, but consumed by none, and dark gray being nutrients which only have one model interacting with them. The network is simplified by hiding metabolite nodes that are not transported in/out by any model at this time point (normally displayed in white color), as well as metabolites not useful for biological interpretation of interactions, including biomass subcomponents (Protein_biosynthesis_e0, RNA_transcription_e0, and DNA_replication_e0) and highly connected ubiquitous metabolites (H2O_e0, H+_e0, Orthophosphate_e0). Minor manual rearrangement was conducted for the expanded network to improve clarity. Five of the six model metanodes are collapsed for clarity, but H. pylori is shown expanded, displaying the environmental exchange reactions. Nutrients of interest, Oxygen, D-Glucose, D-Galactose, (S)-Lactate, and Formate, have been labeled. Part of the L-Proline intracellular pathway has been expanded to exemplify VisANT’s capabilities.