Figure 2. CNV reporter failure does not impact parameter inference.

(A) Flow cytometry of a representative WT population with a persistent one-copy GFP subpopulation, bottom in each panel. FSC-A is forward scatter-area which is a proxy for cell size (x-axis). GFP fluorescence was measured in arbitrary units (a.u.) (y-axis). Hierarchical gating was performed to define the one-copy GFP and two-or-more copy subpopulations (see Methods). (B) Model illustration. $X_A$ is the frequency of ancestor cells in the chemostat;$X_{C^{+}}$, $X_{C^{-}}$ are the frequencies of cells with GAP1 duplications with two or one reporters, respectively, and a selection coefficient $s_C$; $X_B$ is the frequency of cells with other beneficial mutations and a selection coefficient $S_B$. GAP1 duplications form with a rate ${\text{δ}}_C$, other beneficial mutations occur with rate ${\text{δ}}_B$ . At generation 0, only genotypes $C^{-}$ and 𝐴 are present, with frequencies of $X_{C^{-}}=\text{φ}$ and $X_A=1-\text{φ}$. (C) Examples of total CNV proportions (dashed) and reported CNV proportions (solid) for two parameter combinations, both with $s_C=0.15$, $\text{φ}={10}^{-4}$.