A better synthesis: AI-enabled route scouting
The first step is to design an efficient synthesis. The route used by the medicinal chemists who designed the molecule will rarely be appropriate for large-scale manufacture. Speed is of the essence at this stage, and reagents or reactions are often used in the discovery lab that do not readily scale up. It is not uncommon for these syntheses to include 20 or more reaction steps, which can be slow, expensive, and have a very low overall yield.
Lonza’s AI-enabled route scouting service is designed to support our talented process chemists in finding better ways to make complex APIs. Chemists routinely use retrosynthetic analysis to work out what reactions might be used to make a molecule, gradually breaking it down into smaller pieces that, ultimately, are commercially available.
Applying AI tools to the task greatly speeds up this process. It is underpinned by extensive reaction databases, as well as our in-house experience and real-world supply chain database. The service will identify one or more potential synthetic routes that are shorter, higher- yielding, and would be appropriate for larger- scale operation - an essential part of any early phase drug development strategy.
Speed is of the essence: high-throughput experimentation
Optimization is often a slow and laborious process. This is where high-throughput experimentation (HTE) and predictive modelling come into play. These tools provide a method for carrying out many more experiments far more quickly than a human chemist could ever hope to achieve on their own.
HTE allows for the testing of various reaction conditions, reagents, catalysts, and solvents in parallel. The advanced robotic system is capable of running up to 96 different reactions simultaneously and much faster than a lab chemist. Samples can be prepared, and reaction mixtures tested and purified without human intervention. This greatly speeds up the time the reaction screening process takes.
These capabilities play a critical role in early phase development, ensuring the most efficient and scalable route is identified at the outset.
Fit for scale-up: Design2Optimize™
Another piece in the Lonza early phase development toolkit is Design2Optimize™. This platform is designed to enhance the process development and manufacture of small molecule APIs. It is based on an optimized design of experiments (DoE), the statistical approach routinely applied to the optimization of reaction conditions and processes.
Even with an improved synthesis that has far fewer steps, there will still be a multitude of reactions that need to be optimized. Lonza’s Design2Optimize™ was developed along with the Fraunhofer Institute for Industrial Mathematics, enabling drug developers to build predictive models much more quickly. This is even the case if the reaction mechanism is unknown.
It combines physicochemical and statistical models with an optimization loop. This allows chemical processes to be improved with far fewer physical experiments than more traditional statistical methods would require. Importantly, it also generates a digital twin of the process, allowing alternative conditions and scenarios to be tested in silico without further physical experiments. These innovations are core to our early phase drug development approach.