Project Kicked-Off!

1 Apr 2025
Dr Benedetta Bassetti, Dr Jasmine Catlow, Dr Shuyuan Zhang

Academic-Industrial Kick-Off Meeting

On January 22, we kicked off the OptiMed project with a fantastic meeting in Leeds! It was great to bring together all our academic and industrial partners for an engaging and inspiring discussion. We explored key objectives, exchanged ideas, and identified both challenges and opportunities as we move forward.

Our meeting revolved around integrating high-throughput automated screening platforms with advanced machine learning to optimise the synthesis and scale-up of pharmaceutical compounds, associated with insightful discussions about experimental and digital infrastructures, chemical data collection, data-driven approaches, etc.

Some key points are summarised:

  1. Experimentation: Utilisation of flow platforms to rapidly collect data for translation to batch reactors; autonomous process development reporting.

  2. Data: Importance of documents and standardised protocols; importance of collecting metadata; usability of literature data.

  3. Optimisation methodology: Integrating hypotheses and mechanistic studies with optimisations; possibility to benchmark methods.

  4. Rapid transition to manufacturing: Interaction of equipment with data; cost of technology; chemist-friendly interfaces.

See what our PDRAs shared:

Benedetta

In the last few months we welcomed some new pieces of equipment into our laboratory in Leeds, that will be critical for the development of our project: a new Gilson GX-271 liquid handler for the generation of reaction droplets and a Spinsolve benchtop NMR 80 ULTRA for online reaction monitoring. We have been trained, and we are working on integrating them into our systems, as well as in the development of our automated flow platform for reaction screening and process optimisation.

We also started thinking about what type of chemistry we want to explore in the future, and as a starting point we are investigating SNAr reactions, and an amide formation. We conducted some batch experiments, and later a self-optimisation with one of our automated continuous flow platforms.

I was also able to attend the CMAC Open Day recently, and present a poster related to this project. In particular, we presented how we want to apply transfer learning for chemical reaction development. This machine learning approach can be particularly useful for predicting reaction outcomes, optimising reaction conditions, and guiding the design of new reactions, especially when data is limited. In traditional reaction development, chemists often rely on expert knowledge and empirical trial-and-error, which can be resource-intensive and time-consuming. Machine learning, particularly transfer learning, offers a promising alternative by enabling models to generalise across different reactions with limited data.

Jasmine

Shortly after our last blog post, I attended the Leverhulme Research Centre for Functional Materials Design 3rd Biennial Symposium, with a focus on machine learning and data-driven techniques for designing new materials. I especially enjoyed Dr. Nicola Bell’s talk about designing automated lab equipment to make her research on pyrophoric and radioactive materials safer for the researcher. This experience has been very useful for considering how to incorporate more automation into our own research. Since the conference, our group has made several steps towards this goal. We recently hosted engineers from Accelerated Materials, a spin-out company from the Lapkin group. They completed the installation of AMLearn, user-friendly software designed to integrate automation and facilitate reaction optimisation without the need for prior coding experience. We have also been trialling a Biotage V-10 evaporation unit, and investigating the use of Zaiput membranes for inline liquid-liquid separation, as we consider how to accelerate the work-up and purification procedures. Over the next few months, we will be looking to scale up our first set of conditions obtained from reaction optimisation conducted at Leeds.

Mohammed Jeraal and Zara Cheema from Accelerated Materials visiting the Slater Group at the University of Liverpool, Materials Innovation Factory.

Shuyuan

Model-based reaction optimisation offers a promising pathway toward sustainable chemical processes, addressing key challenges such as small datasets, experiment design, and scalability. To achieve this, two questions should be answered: i) how to represent process and model knowledge in a general way, and ii) how to automate model generation. The knowledge graph can be a possible solution, providing a structured approach to managing and leveraging chemical knowledge. Our previous work has validated its effectiveness in constructing chemical process models. The further combined utilisation of model, algorithm, laboratory automation in this project is promising to make a difference in reaction optimisation!

This milestone event marks the beginning of a dynamic collaboration aimed at driving innovation in pharmaceutical development. A huge thanks to our industrial partners—UCB, ChemAI, AstraZeneca, Labman, and RutterDesign—for joining us in this productive meeting. Looking forward, we are thrilled to see where this exciting OptiMed journey will take us.

Stay tuned—more updates are on the way!