Figure 5. Large‐scale parameter scan reveals Pareto optimality between lag times and futile cycling.

• A, B
  Model calculated for randomized sets of protein abundancies, reaction rates, Michaelis constants, allosteric interactions, transcriptional regulation, see Appendix. Each point corresponds to a parameter set that allows exponential growth on both glycolytic and gluconeogenic carbons, as well switching between both conditions. Data are colored according to the total regulation R, i.e., the sum of fold changes in enzyme activities between glycolysis and gluconeogenesis, ${\left(c_i^{\mathit{gly}}/c_i^{\mathit{gng}}\right)}^{{\alpha }_i}$, where $c_i^{\mathit{gly}}$ and $c_i^{\mathit{gng}}$ are protein abundances in glycolysis and gluconeogenesis of protein i and ${\alpha }_i$ the strength of the allosteric regulation. For standard E. coli parameters R = 23. R > 10⁴ are likely unphysiological. Lines indicate Pareto front and are drawn by hand.
• C
  Parameter sets from panels (A) and (B) with low futile cycling highlighted over the background of all parameter sets (gray).