The development of new therapeutic modalities, such as mRNA-based treatments, demands manufacturing platforms that are not only flexible and scalable, but also cost-effective. A persistent challenge in this context is the creation of robust purification processes, typically based on sequential chromatography steps that are complex, time-consuming, and resource-intensive.
To address this bottleneck, a team of researchers from iBB, University College London and LASIGE explored the application of Bayesian optimization as a powerful tool to streamline process development. Focusing on mRNA affinity chromatography, we aimed to enhance the platform’s dynamic binding capacity while significantly reducing the number of experimental iterations required.
The results were remarkable. In just 13 optimization runs, the approach delivered a 7.5-fold increase in binding capacity, achieving 1.8 mgRNA/mL—a substantial improvement over the benchmark. This efficiency demonstrates the strong potential of Bayesian methods to navigate complex solution spaces quickly and effectively.
Importantly, the methodology also integrated model interpretability techniques, enabling us to correlate predicted outcomes with experimental data and gain valuable insights into process behavior. This not only improved trust in the model but also contributed to a deeper understanding of key process parameters.
By embedding Bayesian optimization within a Quality by Design (QbD) framework, this approach represents a significant step forward in automated and data-driven bioprocess development. Its application could extend well beyond mRNA purification, opening the door to broader adoption across the biomanufacturing landscape.
This research contributes to the literature by demonstrating that Bayesian optimization can significantly accelerate the development of mRNA affinity chromatography processes, achieving a 7.5-fold increase in binding capacity within just 13 iterations. By integrating model interpretability techniques, this approach aligns with Quality by Design (QbD) principles, offering a pathway toward more efficient and automated bioprocessing workflows.
Read the full research at: https://doi.org/10.1016/j.seppur.2025.132881