ReactWise partnered with a global top-5 pharmaceutical company to evaluate our software for optimization of biocatalysis systems under realistic experimental constraints. The engagement had two primary objectives: efficiently guide experimental design to achieve the target yield with the fewest experiments possible, while identifying the most influential process parameters.
The study focused on optimizing a biocatalytic reaction across four continuous variables: time, temperature, pH value, and solvent volume percent. Historical results suggested that the optimization would typically require 30–40 experiments to reach the desired yield, so a target of 15 experiments was set for the platform. The team then used a Bayesian optimization workflow to explore the parameter space and rapidly identify the ideal operating regions, while quantifying the importance of each variable to the yield.
The optimization achieved the target yield of 95% in just 12 experiments - a 60–70% reduction compared to similar campaigns. Beyond identifying the most promising operating region, the platform probed the broader chemical space to uncover parameter interactions and relative importance. ReactWise generated clear insights into which of the four process inputs had the strongest effects on yield and where dependencies existed among them. The figure illustrates the relative importance of each input variable.
