Automated Conditional Screening of Escherichia Coli Knockout Mutants in Parallel Adaptive Fed-Batch Cultivations

Abstract

In bioprocess development, the host and the genetic construct for a new biomanufacturing process are selected in the early developmental stages. This decision, made at the screening scale with very limited information about the performance of the selected cell factory in larger reactors, has a major influence on the performance of the final process. To overcome this, scaledown approaches are essential to run screenings that show the real cell factory performance at industrial like conditions. We present a fully automated robotic facility with 24 parallel mini-bioreactors that is operated by a model based adaptive input design framework for the characterization of clone libraries under scale-down conditions. The cultivation operation strategies are computed and continuously refined based on a macro-kinetic growth model that is continuously re-fitted to the available experimental data. The added value of the approach is demonstrated with 24 parallel fed-batch cultivations in a mini-bioreactor system with eight different Escherichia coli strains in triplicate. The 24 fed-batches ran under the desired conditions generating sufficient information to define the fastest growing strain in an environment with varying glucose concentrations similar to industrial scale bioreactors.

Knowledge transfer across cell lines using Hybrid Gaussian Process models with entity embedding vectors

“…With the rise of high-throughput technologies [9], the process development bottleneck has shifted to i) gathering and analysing the generated experimental data; as well as ii) designing the many parallel experimental runs, such that the data of each run is informative. While statistical Design of Experiment methods have come some way towards addressing this challenge [10], these methods largely disregard prior knowledge about the process at hand [11] (in particular when it comes to changes in the cell line), because i) this knowledge is not available in a format that could be explored for experiment design, and ii) though data are produced for similar processes, set-ups for iterative and successive learning (knowledge gathering) from the data are scarce.…”

“…This implies that for every new product a similar number of experiments will be required to understand the process behaviour. In previous articles, the possibility of creating more powerful process models has been explored [12,13], which can be combined with the use of model-based Design of Experiment [14,15,16,17,18,19,20,9] to reduce the experimental effort. However, for a successive acceleration of process development activities novel methods are needed that allow to derive information from a joint analysis of process data generated for different product and further are even capable to predict process behaviour for novel products, for which yet a limited amount of data has been generated.…”