Mechanistic and data driven models for optimisation and troubleshooting cell and gene therapy processes  

The process of lentiviral vector production is complex and presents unique challenges. When developing new products obtaining the optimal titre (best productivity) can be critical to ensure clinical and commercial success, therefore, optimising the plasmid ratios and culture conditions is key. Fast and efficient scale-up of manufacture and control of large-scale bioreactors presents additional challenges.  

The Computer-aided Biology group at OXB has cultivated a unique modelling hub with digital and physical architecture that combines data, models and robotics. This modelling hub allows both data driven and mechanistic models to be run easily by operators.  

We will present the following case studies:  

1. Use of data driven approaches to optimise plasmid ratios based on historic data and associated metadata. 

2. Use of mechanistic models to troubleshoot large scale challenges. 

In Case Study 1 – Advanced DoE Methods, we will discuss the use of data driven approaches to optimise plasmid ratios. Our platform routinely optimises 5-10 factors with the capacity for modelling 100+ variables. Here, we show the generation of a custom DoE via our unique framework, giving optimal space coverage, automatic model diagnostics, automated execution via liquid handler and in silico analysis and reporting. 

In Case Study 2, we will discuss the use of a mechanistic model to troubleshoot large scale challenges. We will present an omics enforced mechanistic model of cell health, metabolism, and vector production. Our model allowed us to highlight cell growth abnormalities, recreating the phenotype at small scale and explore solutions to resolve issues.