Fig. 5: Fitting non-standard microbial growth kinetics with Kinbiont.

Kinbiont can fit microbial growth dynamics with a variety of profiles. A Non-monotonic growth with cell death: Kinbiont fit to an E. coli culture infected by T4 phage²⁰. B Multidimensional fit to biomass, substrate, and ethanol time series from a Kluyveromyces marxianus bioproduction experiment²¹. C Examples of segmented fits for two different alanine-to-glutamate concentration ratios (top: 4, bottom: 1/16). The optimal number of change points was determined using the AICc criterion (Eq. (M13)), evaluated via direct search across multiple segmentation possibilities. D Heatmaps of time-resolved optical density (OD) measurements from diauxic growth experiments (left, data from Ref. ¹⁹). The optimal number of change points for each experimental condition is annotated. Violin plots with embedded box plots (showing the 0.25, 0.50, and 0.75 quantiles; right) display the distribution of mean AICc values for fits with different segment numbers. Conditions that Kinbiont classified as having one change point (8 samples) are contrasted with those requiring two change points (17 samples). See Supplementary Fig. 6 for detailed AICc distributions across replicates. E Mean growth rates for each detected phase are plotted against nutrient composition. Each condition includes 2-4 replicates (see “Methods”); error bars show the replicate standard deviation. Green dots represent the first-phase rate, present in every curve, whereas the purple dots represent the second-phase rate, detected only when a diauxic shift occurs. The dashed black line marks the transition to diauxic behavior, where the change-point analysis (AICc) favors two segments over one.