Figure 3. Comparison with ¹³C-labelling metabolic fluxes in glucose (from refs ¹⁹,²⁰).

Experimental fluxes are compared with the flux calculated with the greedy hypothesis for intermediate and final steps of the adaptation for different knock-outs. The intermediate states along the trajectory have been chosen based on the growth rate of the corresponding ¹³C-labelling experiments used for the validation. Value of this growth rate is reported in each panel. (a) GLU-ppc (40% of recovery with respect to the final growth rate); (b) GLU-tpi (36% of recovery with respect to the final growth rate); (c) GLU-pta (84% of recovery with respect to the final growth rate); (d) GLU-zwf, only experimental data for intermediate steps is available (48% of recovery with respect to the simulated end-point); (e) GLU-pgi (50% of recovery with respect to the final growth rate). In the latter case, the end-points for two different phenotypes are reported (denoted HIGH and LOW). For statistics and comparison with other methods see Supplementary Table S2. The last column reports the key silenced reactions (and pathways) predicted by GRAM and confirmed by the experimental masurements.