A model-theoretical framework to support decisions and accelerate the Design-Build-Test-Learn cycle in Industrial Biotechnology applications

The research objective is to build a model-theoretical framework to accelerate Design-Build-Test-Learn (DBTL) cycles in Industrial Biotechnology applications. The research introduces breakthroughs to connect underlying sciences with lab experiments and process engineering. Synthetic biology offers an extremely promising knowledge basis producing substantial volumes of data and information to assist with the development of custom-made biochemistries. The explosion of data masks poor interactions between upstream science and engineering as the DBTL concept is challenged by poor interactions between upstream science and engineering. The project breaks with practices of the past leading to innovations to reverse-communicate engineering questions back to science. It is rooted in advanced analytics, machine learning, and mathematical optimization capitalizing on rich, and diversified data at the IBISBA infrastructure with access to tools for insilico computing experiments, datasets from experimental labs, facilities to validate innovations, and means to exploit and disseminate research. The first research objective would be to explore supervised and unsupervised machine learning technology experimenting with different technologies, norms, graphs, and exploiting the TasCu ontology of IBISBA. A separate objective will combine space reduction methods, clustering, and visualization to systematically connect observables of Test stages with thermodynamically curated kinetics produced in the parameter space of Design. Research will further explore new discrete-continuous optimization models setting up superstructure schemes that will expand the existing set of degrees of freedom providing for selections of process equipment simultaneously with the reaction pathways. The research will finally convert conventional models on metabolomics to models accounting for engineering aspects such as micro-mixing. The models will be formulated as optimization models to suggest reaction modifications that yield robust choices for reaction and reactors.