Robotics and automation play an important role in biotechnology to handle the ever growing amount of samples and compounds that need to be analyzed. The ability to conduct large screenings with minimal requirements is a key prerequisite in modern bioprocess development. Combined with the emerging sensor and analytics technology these experiments are capable to generate large amounts of data promoting a better understanding, monitor and control of cultivations if handled properly. Finally, we should take process dynamics into account already at early developmental stages in order to speed up times from product to market.
It is therefore necessary to apply mathematical methods for noise filtering, data correlation and model reduction to handle the measurements and extract the contained information. Furthermore, the optimal operation of such complex dynamical and highly parallelized experiments offer new challenges. The chair for bioprocesses at the TU-Berlin investigates novel methods to the design of automated experiments implementing model-based screening, scale up/down, and operation of the robotic facilities. To achieve this, we focus on tractable dynamical models that are tailored to meet the state of information the experiments and allow an efficient and precise model fitting.
Finally, optimal high throughput screening strategies are selected with optimal experimental design methods. Dynamical optimization programs search for the set of parallel experiments which maximize the information content obtainable with the mini-bioreactor parallel setting.
Current activities focus on the development and implementation of scale-down experiments in 24 bioreactors in parallel. Optimal Screening Design for Knock Out strain selection, and Adaptive Structure Models for Yeast 48 parallel cultivations among others.