Developing downstream processes for new drug candidates is costly and traditionally requires extensive laboratory experiments. However, digital tools, particularly mechanistic models, offer an alternative by enabling in silico experimentation, saving time and resources. These models mathematically represent physical and chemical interactions in bioprocessing systems but still require real-world data for validation.
To automate digital twin generation, researchers at Lund University conducted a case study using an ÄKTA Pure chromatography system (Cytiva, Uppsala, Sweden). They developed a workflow that moves from an initial chromatographic setup to a fully calibrated digital model and an optimized operating point. The process involved:
- Defining mathematical equations for the chromatography system.
- Conducting automated experiments to generate necessary calibration data.
- Importing and analyzing data within a Python-based tool called Orbit.
- Calibrating the model with experimental results.
- Using the model to determine optimal process parameters.
- Validating the model by repeating experiments in a physical system.
Orbit, an automation software developed at Lund University, played a key role in integrating laboratory equipment control with mathematical modeling. It enabled seamless data collection, experiment execution, and process optimization. By leveraging nonlinear programming techniques, the researchers successfully designed an efficient chromatography process while minimizing manual intervention.
The study demonstrated the feasibility of automated digital twin creation for chromatography-based purification. However, it was a study on a single chromatography purification step. In a real-world scenario, multiple process steps could benefit from modeling and optimization using this framework. Therefore future work aims to expand this framework to multi-column chromatography and other bioprocessing operations, such as membrane filtration. The researchers emphasize the importance of open communication protocols in laboratory hardware, as these facilitate innovation in modeling and automation.
Bron: www.bioprocessonline.com