(327f) Physical and Data-Driven Modelling of in Vitro Transcription Process

In Vitro transcription (IVT) process is the key part of the RNA production process [1]. Accurately modelling the general IVT process can significantly contribute to developing soft sensors and digital twins for automatically manufacturing RNA products [1, 2]. In this work, we utilised the up-to-date process knowledge and experimental data to build more detailed mechanistic models for predicting the RNA yield and its critical quality attributes (CQA) [1, 2, 3]. Meanwhile, the novel Gaussian process (GP) and global sensitivity analysis (GSA) [4] based machine learning approach was employed to construct fast-solving data-driven models, based on the high-quality data from the built physical models and pure experimentation. In future, the constructed mechanistic and data-driven models will be used to guide process automation and quality by design (QbD) [5] for the RNA manufacturing.

Keywords: Gaussian Process, RNA manufacturing, In Vitro transcription, Global Sensitivity Analysis, Soft Sensor, Digital Twin

[1] Berg D van de, Kis Z, et al. (2021) Quality by Design modelling to support rapid RNA vaccine production against emerging infectious diseases. NPJ Vaccines. 6, 1–10.

[2] Kis Z, Kontoravdi C, et al. (2020) Rapid development and deployment of high volume vaccines for pandemic response. J. Adv. Manuf. Process. 2, e10060.

[3] Kis Z, Kontoravdi C, et al. (2021) Resources, Production Scales and Time Required for Producing RNA Vaccines for the Global Pandemic Demand. Vaccines. 9, 1–14.

[4] Yeardley AS, Bellinghausen S, et al. (2021) Efficient global sensitivity-based model calibration of a high-shear wet granulation process. Chem. Eng. Sci. 238, 116569.

[5] Daniel, S., Kis, Z., Kontoravdi, C., et al. (2022). Quality by Design for enabling RNA platform production processes. Trends in Biotechnology.