Fig. 2: Determinants of recombinase-based differentiation performance.
a, Branching device architecture with different yeast promoters placed on the left or right arm. b, Heat map summarizing differentiation outcomes for 90 strains containing all pairwise combinations of constitutive promoters, ranked from strongest (pTDH3) to weakest (pREV1). Red intensity denotes the fraction of red fluorescent cells; pie charts show representative progeny distributions (black outlines). c, Branching device design incorporating nonsense (NS) DNA (random DNA sequence with no specific structure) to modulate arm length. d, Flow-cytometry analysis of differentiated populations for arm-length ratios from 1:10 to 10:1 (mean ± s.d., n = 8 biological replicates). e, Device design using attP mutants with distinct recombination efficiencies; mutated bases are highlighted in red. f, Differentiation outcomes for seven attP mutant circuits (mean ± s.d., n = 8 biological replicates). g, Summary of five factors (1–5) affecting differentiation output. h, Comparison of progeny ratios for circuits incorporating each factor in g, showing that the first three (starred) significantly expand output distributions (mean ± s.d., n = 8 biological replicates). i, Nine representative circuit designs used to validate the predictive model. The NS DNA was 1,546 bp in length. j, Comparison of model-predicted Venus fractions (lines) with experimental data (dots) from i (mean ± s.d., n = 8 biological replicates).
