Fig 1.
Pictorial representation of the kinetic parameterization pipeline for constructing kinetic models of metabolic networks.
(1A): A set of isotopic labeling data across a range of genetic and/or environmental conditions must be generated or procured. (1B): A stoichiometric model must be constructed. (2): 13C-MFA is performed, and flux ranges are elucidated using the procured isotopic labeling data across all strains from step 1A and the stoichiometric model constructed in step 1B. (3): The flux distributions that were generated in step 2 are used as training data for parameterizing the kinetic model using the stoichiometric model constructed in step 1B and the K-FIT algorithm.
Fig 2.
Reaction and metabolite abbreviations provided in S4 File.
Fig 3.
Reaction and metabolite abbreviations provided in S4 File.
Fig 4.
k-ecoli74 fitness to mutant flux distributions used for kinetic parameterization.
Fig 5.
Comparison of k-ecoli74-predicted flux values with 13C-MFA flux values.
Fig 6.
Number of k-ecoli74-predicted fluxes from each strain used for parameterization falling with one, two, three, or four SDs of the corresponding 13C-MFA value.
Fig 7.
Comparison of mutant strain predicted scaled metabolite concentrations with wild-type metabolite concentration (all values scaled by wild-type absolute metabolite concentration).
(A) Δpgi relative concentration (B) Δrpe relative concentration (C) Δeda relative concentration (D) Δedd relative concentration (E) Δfbp relative concentration (F) Δzwf relative concentration (G) Δgnd relative concentration. Error bars denote range of a single standard deviation from mean scaled concentration value.
Table 1.
SSR value and degrees of freedom (DOF) from 13C-MFA flux elucidation for wild-type and 7 single gene deletion mutant strains using reduced model, SSR value for K-FIT kinetic parameterization using reduced model elucidated fluxes as training data.
Table 2.
Flux predictions using kinetic model parameterized with reduced network flux dataset deviating significantly from k-ecoli74 predictions.
Table 3.
k-ecoli74 metabolite yield predictions under genetic conditions not used for parameterization and comparison with experimental values and values reported for identical strains using the k-ecoli457 model.
Table 4.
A comparison of k-ecoli74 metabolite yield predictions and reduced model metabolite yield predictions under genetic conditions not used for parameterization.
Fig 8.
Comparison of overproducing strain metabolite yield ranges when top three models are used to evaluate target metabolite yields with experimental ranges and k-ecoli457 yield values.
Table 5.
Leave-one-out cross-validation parameterization results.
Fig 9.
Leave-one-out cross-validation comparison with full parameterization.
(A) Predicted cross-validation SSR vs. full parameterization SSR per-strain comparison (B) Predicted cross-validation glucose uptake rate vs. full parameterization glucose uptake rate per-strain comparison.
Fig 10.
Regulatory mechanisms that are dispensable and indispensable to k-ecoli74 fitness.