We developed a workflow to enhance protein purification modeling by exporting protein surface and charge data, utilizing Laplace-Beltrami spectral analysis for feature capture, and computing new local and global descriptors. These descriptors improved the predictions of chromatographic models. Future work includes applying this approach to other complex shapes, like turbine blades, and integrating machine learning for model refinement and interpretability, potentially in collaboration with other ICNAP sprint projects.
| Topic Fields | |
| Published | 2022 |
| Involved Institutes | |
| Project Type | ICNAP Research/Transfer Project |
| Responsibles |
Proteins are biological macromolecules and a central building block of life on our planet. For use in therapeutics, diagnostics or the food industry, proteins need to be purified from complex matrices consisting of thousands of molecules including other proteins. Due to numerous options in chromatography, modeling such purification can substantially reduce experimental efforts during process development and thus safe costs.
Scalar descriptors are one way to capture complex object properties, such as the diameter, overall charge etc. of proteins. However, due to the anisotropic distribution of these properties and the non-spherical shape of proteins, current descriptors fall short of capturing the complex features of proteins like the uneven surface charge distribution.
Here we developed a workflow to i) export protein surfaces along with the corresponding charge distribution in a format amenable for further automated processing, ii) implement Laplace-Beltrami spectral analysis that can capture complex object features, iii) an algorithm to calculate the corresponding local and global descriptors and iv) apply these new descriptors in chromatographic model building (Figure 1). Using the new descriptors (but a fix total number of descriptors), we were able to improve the model predictions of steric mass action (SMA) model iso-therm parameters.
There are several follow-up options to this project, including analyses of other complex objects such as turbine blades as well as advanced model quality analysis and automated/integrated interface between machine learning modelling tools and expert domain knowledge for iterative model improvement and increased interpretability/plausibility. These activities will benefit from an expanded interaction with other ICNAP sprint projects.
Contact us to get in touch! With a membership, you’ll gain full access to all project information and updates.
© Fraunhofer 2025
