Fig. 1.
Experimental and theoretical setup for the characterization of higher-order interactions. Schematic representation of a drug combination plate, where the shaded well in the first row represents the control strain with no drug added and colored wells correspond to single or N-way (up to five-way) combinations from a set of drugs denoted by Xi. The N-way combinations of drugs are represented by wells divided into N identical slices with the colors signifying the drugs in the combination (see 1-drug row for each color). The concentration of each drug is kept the same across single, two-, three-, four-, and five-drug combination experiments. Here, the experimental setup is simplified for illustration, but in actuality, (1) we filled all wells with bacteria and drug combinations to obtain replicate measurements (see “Experimental details”), and (2) we used multiple 96-well plates for each five-drug combination. From these experiments, fitness of bacteria (w) in the presence of drug combination (D) is assessed by the relative growth rate with respect to the no-drug control (WT). In the figure, schematics for the net N-way interaction include all possible lower-order connections, whereas an emergent interaction schematic connects all N drugs (such as dyad and triad for two- and three-drug combinations, respectively)