Managing uncertainty in metabolic network structure and improving predictions using EnsembleFBA
Fig 8
EnsembleFBA predicts unique essential gene targets in six Streptoccocus species.
A. We reconstructed ensembles of 21 GENREs for each of six Streptococcus species based on draft GENREs from the Model SEED and 25 growth conditions. We identified genes in all six species which potentially interact with small molecules from the DrugBank database, and used the ensembles to predict the essentiality of those genes. We found 261 small molecules with potential binding to essential gene products. B. Many small molecules interact with an essential gene in only one species, while a core set of 44 small molecules interact with essential genes in all six species. C. Distribution of small molecule interactions with essential genes, unique and conserved among the six species. Note that 44 small molecules interact with essential genes in all six species. S. equinus is predicted to have essential genes uniquely interact with 43 small molecules, while S. pneumoniae is predicted to not have any essential genes which interact with unique small molecules. D. Subsystem enrichment among essential reactions by species. We predicted reaction essentiality for all six species in rich media and then calculated a p-value indicating the likelihood of observing each subsystem among the essential reactions given the total number of reactions associated with that subsystem. For clarity we display the–log(p-value), where darker colors indicate greater enrichment (i.e. a disproportionate number of reactions in that subsystem are predicted to be essential). Note that some subsystems are enriched among essential reactions in all six species (e.g. Peptidoglycan biosynthesis) while others are uniquely enriched in a specific species (e.g. Phenylalanine biosynthesis in S. mitis).