#25 · Systems biology

Flux models reveal how biomass and network topology shape cellular metabolic states

The dominant thread is mass-balance and network-constrained metabolic modeling, applied at scales from single-cell transcriptomes to bioreactors and proteomes. A genome-scale 13C-MFA study of E. coli shows that flux estimates shift with protein/RNA ratio changes of 2-3%, with particular sensitivity in one-carbon metabolism, which makes biomass formulation a direct source of model error. An FBA-based ROS model in augmented iJO1366 reproduces bimodal H2O2 killing as an emergent property of network topology. FluxSprout builds metabolic network ensembles from scRNA-seq of 10,090 zebrafish cells using Zebrafish-GEM and audits annotation, loop, and specificity. A metabolism-informed learning model embeds mass balances for CHO mAb bioprocesses to infer intracellular fluxes and forecast extracellular concentrations during transfer from fed-batch to continuous operation. A compilation of 300 absolute yeast proteomes links subcellular protein allocation to growth rate. What is new is the use of stoichiometric constraints to carry inference from single-cell and bioprocess data to flux-level predictions.