BioSolve Process is the biopharmaceutical industry’s most powerful software for bioprocess analysis. It enables users to investigate a wide range of process scenarios and parameters, and make informed and targeted process design decisions. However, due to the complexity of the bioprocess models and the large decision space, these simulations can be labour- and time-intensive.
This challenge motivates the use of optimisation techniques, such as Bayesian optimisation, that require only a few evaluations to arrive at near optimal solutions. BioSolve Process functionality has therefore been extended to incorporate a Bayesian optimisation tool that automates the selection of the most efficient process parameters, based on pre-defined user criteria. This optimisation tool is customised to tune a combination of categorical and discrete variables, subject to black-box constraints and conflicting objectives.
Firstly, a case study based on optimising upstream decisions over a range of target throughputs for a traditional monoclonal antibody (mAb) process was used to assess the performance of the optimisation tool. The upstream decisions considered included choosing between stainless-steel and single-use bioreactors, the number of installed vessels, and how many bioreactors were pooled together for further downstream processing. Compared to manual full enumeration, the optimisation tool showed a substantial reduction in evaluations.
A second case study incorporated downstream decisions. In particular, the optimisation tool was able to select between conventional or pre-packed chromatography columns, and from a set of Protein A chromatography resins on offer. Finally, the tool was used to estimate the set of trade-off solutions (Pareto front) for multi-objective problems for the entire mAb process train with the aim of minimising cost of goods (CoGs) in combination with capital expenditure, water usage, and process mass intensity (PMI).
The results of these case studies identified scenarios in which certain process decisions were attractive. In addition, they demonstrated the efficiency of the optimisation tool in presenting users with good solutions to their decision problems through BioSolve Process.

Extending BioSolve with multi-objective Bayesian optimization for automated tuning of upstream and downstream decisions in a mAb process

ACS Fall 2020 Virtual Meeting

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.