A systematic framework for biomolecular system identification

Unwieldy challenges, such as emergent behaviour or lack of compositionality, hamper the rational engineering of synthetic biomolecular systems. The use of mathematical models would address many of these challenges, yet current practices in synthetic biology make obtaining them time and resource intensive. More importantly, in many cases, the process of obtaining data for mathematical models is guided by intuition rather than rigorous modelling requirements. To make better use of the available resources, this tutorial proposes an end-to-end framework for biomolecular system identification. Given a biomolecular system inside a biological organism, the framework leverages on system identification techniques to automate the initial proposition of candidate models for the system of interest. Then, statistical methods guide the optimal design of experiments that enable discrimination and calibration of the most plausible model candidates for the underlying system. We foresee that, by establishing a closed-loop between datagathering and model identification, the outlined approach will accelerate and standardise computational modelling of biomolecular systems.