Optimizing Bioprocess Design and Workflow Orchestration

Inefficient experimental design
The traditional “one-factor-at-a-time” (OFAT) testing approach, where one parameter is changed while others are held constant, makes perfect sense in the classroom. Yet in the real world of complex bioprocesses, this approach is both inefficient and limited, leading to fragile experimental methods that can be difficult to transfer to biomanufacturing.¹  Design of Experiments (DOE) offers a systematic alternative by testing multiple parameters simultaneously, but executing and tracking these complex experimental designs across teams requires workflow orchestration not only manage the coordination but also facilitate DOE execution through automation. 

Alberto Pascual, an informatics product portfolio leader at Revvity Signals, uses coffee as an analogy: “To achieve the best cup of coffee, you need to consider parameters like temperature, pressure, and coffee beans. Using an OFAT approach, you’d need to make perhaps 1,000 cups of coffee to try all possible combinations of those parameters, which just isn’t feasible.”
In bioprocessing, while the parameters differ from those used to make a cup of coffee, the core issue of exponential complexity remains the same. In addition, since biological processes are multidimensional, changing one parameter often changes something else. Neha Mishra, a senior bioproduction scientist at Revvity Discovery, explains “In cell-line development, the process from vector construct design to an isolated clone can take 12–17 weeks; it’s quite a long process. There are a lot of different parameters to consider.”

Data access
To make full use of the data from each precious experiment performed in the lab, notes and data need to be carefully recorded in retrievable way. The FAIR principles, defined in 2016, promote the Findability, Accessibility, Interoperability, and Reusability of data in research. ²
In the labs of the past, paper lab books were used to record experimental data, which meant having to search through pages and pages of records to locate the data you need. Mishra’s team switched to using electronic lab notebooks (ELNs), which made data location easier, but it still requires an understanding of which ELN contains the experimental data you’re looking for, since ELNs are very experiment centric. “It would be very helpful to have an automated process that collates data for a particular experiment, so that it’s easy to access,” stresses Mishra.  
Locating information from previous experiments is a huge problem in today’s labs, since the volume of data generated can be so vast. For example, it has been estimated that 85% of health research data is wasted or underutilized.³  When data can’t be easily found, the simplest path is often to repeat the experiment, absorbing time, money, and resources unnecessarily.

Data analysis
Beyond experimental design, data analysis remains one of the major bottlenecks in cell-line development. “We need to analyze cell titers and product quality of a lot of different clones at the same time.” explains Mishra, “That means we have massive quantities of data to analyze so that we can pick the right clone, and that's time-consuming.” 
Similarly, considerable effort is spent transferring data from one system to another, such as from one instrument to a USB stick to another instrument. “On average, daily, we could easily spend maybe an hour or two doing just that,” stresses Mishra, “We have different cell counters and different instruments for measuring product quality or titer, and everything has an individual setup; none of the instruments talk to each other.” 
Apart from the time cost, there is also the risk of introducing errors when transferring data between instruments and analysis spreadsheets: “Sometimes it takes time to realize that an error was introduced, and you have to go back and dig for the data again. This is why we store all our data for at least six months.”

Data visualization
In common with most bioprocess research, cell-line development produces significant data volumes. Process scientists compare graphs created by multiple instruments and software applications to assess the progress and results of each experiment. “Another time saver,” adds Mishra, “would be to have a way of getting a visual overview of what an experiment is actually telling you.” 

Regulatory compliance
To reduce unwanted variability, bioprocess experiments follow standard operating procedures (SOPs), with a series of predefined steps to ensure that all researchers are performing the experiments in the same way. Mishra explains that in cell-line development, multiple separate SOPs apply, such as one for transfection, one for titer analysis, etc. This means that onboarding new staff takes time, as there is a need to ensure they understand which SOPs to follow, and in which order. 
In addition, with many different reagents, consumables and samples, traceability and sample ID tracking are crucial. Keeping track of all this information is time consuming, and Mishra estimates that she spends one to two hours per week checking that critical audit data have been properly recorded. To complicate things further, different lab instruments may have their own unique sample ID systems, and manually mapping these from system to system without risking data integrity presents a significant challenge.

Task assignment and tracking
In small organizations, one team might perform all aspects of the R&D and product quality analysis. However, in larger biopharmaceutical companies, the workflow is more compartmentalized, and teams need to know when to expect a task to arrive on their desk. Asking people or sending emails to assign and track tasks can cause issues that result in wasted resources. “One person could have had assigned somebody a task,” explains Mishra, “But then I come in and I had no idea that the task has been assigned, and I start doing it myself. Sometimes this might mean that we are doing double the work.” On the other end of the scale, miscommunication can lead to deadlines being missed. Mishra emphasizes that: “Even if a spreadsheet is used to keep a record of what each individual in the team is doing, this only works if everyone checks and updates it on a daily basis, which is time consuming.”