Fig. 4: Our machine learning approach successfully extracted 15 sensors, each with distinct malathion response curves.

a A map of the plasmid, pBHVK, used to construct the library. The plasmid contains a kanamycin resistance gene as well as a fast-folding sfGFP gene. b Average per cell sfGFP signal at 0.37 μM (left) and 1.83 μM (right) malathion normalized by signal at 0.0 μM malathion is shown for all 15 engineered strains. c Transfer curves (or dose-response curves) for each strain are depicted with markers and their fit to Hill equation kinetics are given by solid lines. The Hill equation parameters are given in Table 1. The promoter sequences corresponding to each reporter and time points for each transfer curve are given in Supplementary Tables 2 and 4, respectively. The transfer curves are plotted at the time point of maximal fold change of the 2.24 μM response with respect to the 0 μM response. The error bars represent the standard deviation from the mean across three biological replicates. d Transfer curve parameters for the dose-responses depicted in c. The error bars represent the standard deviation from the mean across three biological replicates. Source data are provided as a Source Data file.