Hybrid Model of RNA Bioprocessing
When asked about how the model could cover such a large scale, Xie replied that \”it is rooted in an integration of mechanistic modelling and ML statistical techniques, which allows the models to provide a comprehensive and nuanced understand of various aspects of RNA, and related processes while quantifying uncertainty due to limited information.\”
She explains, for example, that \”the mechanistic element of the model captures intricate chemical and physical properties at the subatomic level which supports a deeper understanding of the biological processes.\” The machine-learning component can capture patterns from complex datasets, such as molecular simulators and time-course fluorescence microscopy data, and learn relationships that may not be explicitly described by existing mechanistic modeling.
The hybrid model developed by the Northeastern University team promises to have many commercial advantages in the production and development of monoclonal antibody, cell and gene therapy, and mRNA vaccinations. As Xie explains, \”It is a way to advance the knowledge of RNA production mechanisms and guide simultaneous control strategies at different levels such as RNA sequencing and specification of critical quality characteristics, with fewer experiments.\”
Source:
