Comparison of CFU-based and luminescence-based rates of change.

For each drug, we generated 2000 bootstrapped datasets by resampling time-series CFU and light-intensity data with replacement and fitted an exponential function to each bootstrap replicate to obtain distributions of rates. Panel (a) shows these distributions for 20 drug–concentration assays across 19 antibiotics; panel (b) shows the antimicrobial peptide pexiganan at and . The distribution of ψCFU is shown in blue (lower half of each violin, diamond). The distribution of ψI is shown in orange (upper half, triangle). Green distributions ( triangle) represent luminescence-based rates calculated from data starting at the first peak onward. Red distributions show volume-adjusted luminescence rates (ψJ; pentagon). Vertical lines mark the 95 % confidence intervals. Asterisk (*) or letters (n.s.) indicate whether CFU-based rates differ significantly from the corresponding luminescence-based rate or not (see Methods), with color coding matching the respective luminescence-based distribution. Wide confidence intervals for pexiganan reflect biphasic killing, steep curves, and noisy CFU data.

Density distributions (2.5–97.5% percentile range) of pooled cell volumes acquired by microscopy imaging, shown (a) before and (b) after 2 h of antibiotic treatment.

Boxes indicate the 25–75 % interquartile range, and vertical bars mark the mean. Significance (*) was assessed by bootstrapping cell volumes 200 times with replacement for each replicate (image) and treatment, pooling the bootstrapped volumes by treatment, and comparing the resulting 95% confidence interval of the mean to that of the untreated control (control_2h). Carbapenems tend to deform cells into a lemon-like shape (Figure S30), resulting in poor fitting quality since our algorithm assumes a cylindrical geometry.

Simulations based on the filamentation model quantifying how changes in division rate due to treatment (Δλ) and death rate (δ) influence the rate of change of population size (ψB, blue), the rate of change of light intensity (ψI, orange), and the rate of change of light intensity when the first 2 hours of data are excluded (, green).

These illustrative simulations were conducted using an initial division rate λ0 = 1.5 h−1. More details and all parameter values can be found in Appendix S3.