Figure 5. Sampling effects on the ranking of drug pairs.

Both panels show correlations between the a priori estimated $c$-rank ($x$-axis) of the 125 Gaussian JDFEs and their a posteriori rank ($y$-axis) based on the risk of collateral resistance observed in simulations (same data as the $y$-axis in Figure 4C for $N{U}_b=0.46$). (A) The $c$ statistic is estimated using the Luria-Delbrück method (see text for details). Cutoff for sampling mutations is $0.5\sigma$, where σ is the standard deviation of the JDFE in the home environment. See Figure 5—figure supplement 1 for other cutoff values. (B) The $c$ statistic is estimated using the barcode lineage tracking method with $N={10}^6$ and $U={10}^{-4}$ (see text and Materials and methods for details). $P\S lt;{10}^{-6}$ for all regressions.

Figure 5—figure supplement 1. Same as Figure 5A, but with different thresholds for sampling.

Same as Figure 5A, but with different thresholds for sampling mutations, as indicated above each panel (σ is the standard deviation of the JDFE in the home environment). See Materials and methods for details. $P\S lt;{10}^{-6}$ for all regressions.