The problem
Designing genetic or protein networks that satisfy a set of behavioural specifications is one of the main challenges of synthetic biology, and model-based design is a natural choice for it.
The paper considers how to tune the parameters of a stochastic model so that one or more behavioural goals hold.
The approach
The goals are specified as Signal Temporal Logic formulae, and the aim is a parameter set that makes their satisfaction probability as large as possible. This is a multi-objective optimisation problem.
It is solved with an optimisation scheme that combines satisfaction probability and the average robustness of the STL properties, using state-of-the-art multi-objective optimisation routines.