There are some challenges with DOEs that are commonly encountered in bioprocessing. DOEs enable an exploration of a number of variables to identify not only the main factors that impact strain performance, but also interactions between factors that impact strain performance. However, beyond screening DOE designs which only identify the main factors due to a small number of conditions, bioreactor capacity is often a barrier to performing fermentation DOEs. Larger designs, which can identify interactions between variables and curvature in response within a variable, can require many test conditions. If capacity constraints are an issue, these large studies will take a long time to complete. They risk either getting trimmed down and run with insufficient replicates leading to inconclusive results or not being prioritized against other types of work.
On many fermentation platforms, defining parameters for multiple fermentation conditions requires arduous manual programming, which can be time consuming and prone to user programming errors. Such errors, which can invalidate experimental conditions, complicate interpretation of experiments such as DOEs where successful execution of all conditions is required to estimate interactions between different variables. In contrast, Culture’s proprietary recipe based programming simplifies input of multiple fermentation parameters in an accurate manner.
There are some best practices that can be applied to maximize the value of DOEs for bioprocessing. By applying these best practices, the most appropriate experimental designs and conditions can be selected to build performance models.
Bioprocesses are an interaction between the organism and the process, and are therefore also somewhat dependent on the physiology of the organism. This can be leveraged by performing experiments in order to narrow down the relevant process design space. For example, an organism may only have growth rates that are acceptable for a process within a defined temperature range. Therefore, range finding experiments, institutional knowledge or precedence in the literature can be used to narrow the number of conditions in an experiment.
Many factors in a bioprocess may influence others. In some cases, a setpoint of one condition may impact another to the extent where it is no longer relevant. By accounting for these interactions in experimental designs, confounding results can be minimized. For example, a low temperature setpoint may reduce growth rates to an extent where a fixed feed regime may result in overfeeding and poor strain performance. In this example, implementing a dynamic feed scheme could ensure appropriate feed rates over a range of temperature setpoints.
An assessment of the goal and stage of a project to identify the relevant design can also be helpful. For example, when building early strains, inserting a heterologous pathway into an organism for a screening DOE may be sufficient in order to identify the main factors that impact performance of a strain. For a more mature project, where a process is being fine tuned, it may be more appropriate to design a larger DOE. By accounting for the goals of a project at the time and in the future the most appropriate DOE design can be implemented.
Data generated in bioprocesses will have some level of variation between replicates both within and between an experiment. It is important to have an assessment of this variability before performing a DOE in order to structure replicates to generate statistically significant results. A well-designed replicate structure can maximize insights derived from a DOE. In addition, bioprocess experiments can generate very large amounts of data. Experiments should be designed to take appropriate measurements at relevant time points in order to make the desired conclusions.
A DOE is an experiment type that can be performed at Culture. DOEs can be designed using standard statistical software and submitted in an experimental plan to Culture. The Culture team uses their proprietary recipe based programming system to program the reactors. The system can use a single recipe to dictate the process parameters for any number of reactors. This is coupled with a tuning and triggers table, where individual process parameters can be specified to vary within the recipe to be implemented in different reactors in the run. This results in rapid execution and high operational success rates of experiments incorporating a large number of conditions.
There are certain factors that differentiate performing a DOE experiment at Culture from running it using internal bioreactor capacity. These factors each contribute to high success rates and data quality, enabling statistically meaningful conclusions from DOEs.
Culture’s bioreactor capacity allows larger DOE designs to be executed experimentally with sufficient replication in order to generate statistically significant results. Even large DOEs can be performed in a single experiment, or rapidly in a series of experiments. This results in accelerated experimental timelines in order to gain meaningful insights into factors that impact performance in a bioprocess.
Many features of Culture’s platform facilitate the generation of high quality data, resulting in statistically significant results that can then be used to build predictive models of performance. There are several hardware precision control features of Culture’s bioreactors, including gases delivered via mass flow controllers and up to 5 feeds delivered via peristaltic pump with scale feedback control. Published mass balance data on every run also ensures quality data. In addition, Culture’s quality guarantee ensures that every condition will be successfully executed as experimentally planned.