Anticipated cell lines selection in bioprocess scale-up through machine learning on metabolomics dynamics

The development of biopharmaceutical therapeutics, such as monoclonal antibodies, requires the testing of several cell lines at different development scales and the selection of the high performing cell lines which allow meeting the desired quality attributes of the product. In this context, data analytics, which is extremely useful for a better process understanding and a faster scale-up, can be used to understand the relation between biological information, such as cell metabolism, and process productivity. This study shows that monoclonal antibodies end-point titer can be estimated in the early stages of the industrial product development for cell line selection using information on cell metabolism dynamics. This allows the anticipated identification of the high-performing cell lines, and a better understanding of the relationships between the time evolution of both the metabolic information and the product titer.