Cell and gene therapy manufacturing is biologically complex and technically demanding. Processes often involve living cells, viral vectors, sensitive materials, and multiple interacting process parameters. Small changes in cell culture conditions, transfection performance, metabolite levels, or harvest timing can influence productivity and final product quality.
Despite this complexity, many workflows still rely on manual operations, offline sampling, delayed analytical results, and fragmented data handling. This can limit process understanding and make it difficult to apply Quality by Design principles effectively.
PAT and automation help address these limitations. Automated sampling enables more frequent and consistent data generation. Inline and offline analytics provide deeper visibility into process behavior. Digital platforms make it possible to collect, structure, and interpret data from multiple instruments. When these elements are connected, they create the foundation for better process understanding and, ultimately, more advanced process control.
Dr. Shaeri summarized this need through three essential elements: sensing, data, and control. Manufacturers need reliable analytical technologies to monitor the process, structured data to understand process behavior, and automation infrastructure to support timely action.
Cell and Gene Therapy Catapult supports the development and industrialization of advanced therapies by helping translate scientific innovation into practical, scalable manufacturing technologies. With R&D and GMP manufacturing capabilities in the United Kingdom, CGT Catapult provides an environment where new technologies can be evaluated and integrated under realistic development conditions.
In his presentation, Dr. Shaeri introduced work from CGT Catapult’s Scale Enabling Technology team, which focuses on areas including digital innovation, automation, gene therapy, cell therapy, and sustainability. The project presented during the seminar sits directly at the intersection of automation and digitalization.
The aim was not only to automate individual tasks, but to build a connected platform that can generate high-quality process data, support modeling, and enable more advanced approaches to process monitoring and control.
A central theme of the presentation was the development of an integrated PAT Lab platform designed to support Quality by Design. QbD requires a strong data foundation. To understand the relationship between critical process parameters and critical quality attributes, process developers need frequent, reliable, and contextualized data.
The platform presented by Dr. Shaeri combines automated sampling, analytical technologies, centralized data capture, software integration, modeling tools, and feedback control. In this setup, automated sampling plays a key role by enabling high-frequency and consistent sample collection from bioprocesses.
The project integrated multiple technologies, including analytical instruments for cell and process monitoring, Raman spectroscopy for inline metabolite monitoring, and digital tools for data handling and modeling. Lucullus was used as the central platform for data capture and process control, while Numera supported automated sampling as part of the integrated workflow.
This type of connected infrastructure is important because individual instruments alone do not create process understanding. The real value comes from integrating data across the process and using it to support analysis, modeling, and decision-making.
The work presented by Dr. Shaeri was demonstrated using an established AAV process as a model system. Viral vector manufacturing is a highly relevant use case for automation and PAT because it involves complex biological interactions and a strong need for improved productivity, quality, and consistency.
In the project, the team aimed to integrate automated sampling, PAT tools, process data, and modeling into a single workflow. The goal was to generate high-quality data automatically, monitor critical parameters more closely, and move toward more advanced control strategies.
Before biological runs were performed, the team conducted site acceptance testing to evaluate the integrated platform. This included testing manual and automated modes of operation, checking dilution functionality, and comparing analytical outputs from different operating modes. This step was important to confirm that the platform could generate reliable data before being used in process-development experiments.
The presentation made clear that successful PAT implementation requires careful preparation. Hardware, software, data interfaces, sample handling, calibration, and user workflows must all function reliably. Without this operational foundation, the quality of the resulting data can be compromised.
One of the major advantages of automated sampling is the ability to generate richer datasets. In the presented work, automated sampling enabled many more process time points than would typically be practical with manual sampling alone.
This matters because biological processes are dynamic. Important changes can happen between traditional manual sampling points. More frequent sampling gives scientists and engineers a clearer view of how the process evolves over time.
However, Dr. Shaeri also emphasized an important lesson: more data are not automatically better. Data quality, reproducibility, and consistency remain essential. In the project, some modeling results were limited by variability between replicates, showing that strong experimental design and robust execution are just as important as high-frequency data generation.
This is a valuable message for the broader field. Digitalization and automation can generate more data, but useful process understanding depends on reliable, well-structured, and meaningful data.
To explore the process design space efficiently, the team used a design-of-experiments approach supported by Latin hypercube sampling. This helped cover a multidimensional process space while keeping the number of experimental conditions manageable.
This approach is particularly useful in cell and gene therapy development, where many factors may interact with one another. Instead of changing one variable at a time, advanced experimental designs can help identify relationships between parameters and support more robust process optimization.
The project also included model development using process data generated through the integrated platform. Modeling was used to compare experimental conditions and predict process outcomes. While the models showed useful potential, the presentation also highlighted that model accuracy depends strongly on data quality and process reproducibility.
This is especially relevant as the industry moves toward digital twins, AI-supported development, and adaptive process control. Advanced models require strong, contextualized, and reproducible datasets. Automation and PAT can provide this foundation, but they must be implemented with scientific and operational discipline.
Another key topic in the presentation was the transition from process monitoring to active process control. PAT is often first used to observe and understand a process. The next step is to use PAT signals to control the process in real time.
Dr. Shaeri presented an example of PID-controlled feeding for metabolite regulation. Raman spectroscopy was used for inline monitoring of glucose and lactate, while the control system supported automated feeding based on defined set points.
This type of feedback control is highly relevant for advanced therapy manufacturing. Cell culture performance can be strongly influenced by metabolic conditions. Maintaining glucose, lactate, and other key parameters within defined ranges can support more consistent process performance.
More importantly, this approach reflects the broader goal of Quality by Design. QbD is not only about analyzing quality after a process has finished. It is about designing processes that can be monitored, understood, and controlled to deliver consistent quality.
Dr. Shaeri’s presentation also provided practical lessons for teams planning to implement automation and PAT.
First, integrated systems introduce operational complexity. Device communication, data transfer, maintenance, and troubleshooting must be planned carefully.
Second, software compatibility and connectivity require continuous attention. Automated platforms often depend on multiple instruments, software tools, and data interfaces, which must remain aligned.
Third, operational readiness should be separated from process optimization wherever possible. Testing and stabilizing the platform before running key biological experiments helps reduce variability and improves data reliability.
Finally, expectations must be realistic. Automation and PAT can provide major benefits, but implementation is stepwise. Success depends on technology readiness, process understanding, good experimental design, and collaboration between scientific, engineering, automation, and data teams.
The broader message of Dr. Shaeri’s talk is clear: the future of cell and gene therapy manufacturing will depend on better integration of biology, automation, analytics, and data science.
More connected and automated workflows can help developers generate stronger datasets, understand process behavior more deeply, and move toward active process control. For viral vector manufacturing and other advanced therapy applications, this can support greater robustness, improved scalability, and more efficient process development.
The PAT Lab platform presented by CGT Catapult provides a practical example of how these technologies can be brought together. It shows both the opportunities and the implementation realities of next-generation automation in advanced therapy development.
We would like to thank Dr. Mohsen Shaeri and the Cell and Gene Therapy Catapult team for sharing their expertise with the Securecell Learn & Connect community.
Watch the full recording to learn more about how automation, PAT implementation, data integration, and Quality by Design can support the next generation of cell and gene therapy manufacturing.