Fig. 3: Summary of the active workflow applied to the PPP network.
From: Timed batch inputs unlock substantially higher yields for enzymatic cascades
a, Training the model. Left: the inflow profile of the first iteration with the flow rates of each enzyme and substrate; the dashed black line represents the total flow rate. Right: data as measured using HPLC (black triangles): AMP, ADP, ATP and NADP. The convergence of the training process to this data is given by the mean (thick coloured line); the uncertainty of the fit is visualized by shaded area (s.d. of 2) utilizing the 25 best fits. For the flow per species, the actual concentrations of enzyme species are shown in Supplementary Fig. 8. b, Testing the model. Left: the 25 best fits from iteration one are used to predict the OED experiment performed for iteration two. Right: predictions of the time series data are visualized in an identical manner to a.
