Fig. 4

Non-numerical multi-objective optimization. Builders of intricate structures before the scientific era were often faced with the need to play a large number of parameters that were not amenable to the calculation tools available at the time. Architects like Antoni Gaudi (1852–1926) figured out ways to solve the problem by making string models of the building or building parts (a) in which weights were hanged at critical places for revealing the effect of local structures on the geometry of the whole object. b Uncertainties on the best combination of enzymatic steps (1–5) for converting a substrate into a product (Z) include inter alia reaching a suitable level of transcription (the function of the promoter P and the regulator R) and adequate intergenic regions (IGR) for ensuring the necessary stoichiometry in protein production, as well as mRNA stability and termination (T). Sequence diversification at such regulatory points and selective pressure to increase production of Z allows exploration of the solution space until an optimum is reached