Fig. 3.

Prediction of chemosensory diversity from Tar/Tsr ratio variation, and establishing causality. (a) (Top) Black points represent the measured coefficient of variation (CV) of the the half-maximal concentration $K_{1/2}$ as a function of OD in rich media, obtained from the single-cell dose-response experiments in Figure 1c. Blue points represent the predicted CV of $K_{1/2}$ as a function of OD in rich media, derived through the mixed-species MWC model using single-cell receptor ratio data from Figure 2d as an input. Shaded areas indicate 95% confidence intervals obtained through bootstrap resampling. (Bottom) Same as the top panel, but for cells grown in minimal media. The CV was either measured from the single-cell dose-response experiments in Figure 1d (black points) or predicted from the mixed-species MWC model using single-cell receptor ratio data from Figure 2d as input (blue points). The total number of MWC receptors (N_total) was fixed to 32 for all rich media experiments and 100 for all minimal media experiments. (b) Histograms of the Tar/Tsr chemoreceptor ratio measured via fluorescence microscopy for wild-type cells (WT population; replotted OD = 0.80 data from Figure 2d) and for a population with artificially increased Tar/Tsr ratio variability (high variability population). Solid lines represent the histograms of all Tar/Tsr ratios from four independent biological replicates (WT population) or a single experiment (high variability population). Shaded areas indicate 95% confidence intervals obtained through bootstrap resampling. (c) Histograms of L-serine $K_{1/2}$ values obtained from single-cell dose-response curves for wild-type cells (WT population; replotted OD = 0.80 data from figure 1c) or cells with artificially increased Tar/Tsr ratio variability (high variability population). Solid lines represent the histograms from a single experiment, with shaded areas indicating 95% confidence intervals obtained through bootstrap resampling. Inset: Measured CV of $K_{1/2}$ for WT and high variability populations (solid-colored bars), along with predicted CV of $K_{1/2}$ obtained from the mixed-species MWC model using the receptor ratio distributions in Figure 3c as input (dashed-colored bars) and a fixed total number of MWC receptors (N_total = 32). Error bars indicate 95% confidence intervals obtained through bootstrap resampling. The means are statistically significant different, as determined by a two-sample t-test (p = 2.9×10⁻¹²).