Fig. 3: Phenomenological modelling of different choices of control input.
From: Genetic controllers for enhancing the evolutionary longevity of synthetic gene circuits in bacteria
a A simple schematic showing the three controller architectures: intra-circuit control (red), growth-based control (orange) and population-based control (lilac). Each output comes from a different layer of the combined multi-scale model (circuit model, host model and population model). See key in Fig. 2a for symbol meanings. b Time-series of population-wide output P over time for an open-loop system (black, dashed, wA = 4.0 mc min−1) and representative control systems of equivalent initial output P0. (Intra-circuit: red, wA = 103 mc min−1, kA = 4.5 × 102 mc. Growth-based: orange, wA = 87 mc min−1, kλ = 6.3 × 10−2 min−1. Population-based: lilac, wA = 16 mc min−1, kP = 5.2 × 108 mc.). c, d A large number of designs were generated by varying the maximal transcription rate ωA and control parameters ku. Against the initial output P0, we plot the percentage change in (c) τ±10 and (d) τ50 vs an open-loop system of equal initial output. Points marked with an X correspond to the time-series plots in (b). These were selected as points on the Pareto fronts simultaneously optimising P0, τ50 and τ±10, with initial output P0 closest to 2 × 109 molecules. e, f For circuits corresponding to trajectories in (b), (e) maximum protein production per cell pA and (f) maximum growth rates λ across the first day (solid) and the day where τ50 is reached (grey outline) for each mutation state. Lighter squares represent less functional strains. The horizontal line represents a non-functional strain. Simulation results are provided as a Source Data file.
