AI researchers and Orion boost drug development with a new virtual laboratory
Innovative virtual laboratories could help, for example, medicinal chemists in optimising chemical reactions. A whole new pharmaceutical ecosystem is emerging around AI in Finland. Lue tämä suomeksi
Image: Mostphotos
This is what it’s all about
The goal is to use new AI methods to support early-stage drug development.
The project will conduct basic research as a step towards creating a new pharmaceutical ecosystem in Finland.
The research will increase understanding of how AI can help pharmaceutical companies through the idea of an innovative virtual laboratory.
Useful advice from AI
When scientists begin to develop a new drug in a laboratory, they must explore the unknown, as they cannot know for certain what molecule will have the desired effect or how the necessary chemical reactions can be achieved. What if artificial intelligence could help them on the way?
This is the goal pursued by the University of Helsinki, Aalto University and the pharmaceutical company Orion in the project Virtual Laboratories for Pharmaceutical R&D. Even a small increase in the speed and accuracy of drug development is significant, as the whole process may easily take up to 15 years.
“If we can make the process more efficient, the benefits are obvious,” says project coordinator and Associate Professor of Computer Science Arto Klami.
AI could, for example, advise scientists on what to do next: should they conduct tests in the laboratory or perform more simulations? And what conditions would be optimal for a specific chemical reaction? Scientists will continue to make decisions, but AI can help them manage the process.
“It’s about AI working alongside scientists,” states Klami.
“We’d like to see similar AI methods becoming common in as many scholarly fields as possible.”
Research supports the creation of a pharmaceutical environment
The two-year project received about €500,000 of Business Finland’s Co-Research funding. The project is linked to Orion’s Veturi programme, which is building a Finnish pharmaceutical research ecosystem using AI. If successful, the new network will have all the skills and data required for drug development projects.
“It will also be an interesting environment for other pharmaceutical companies conducting research in Finland and elsewhere,” says Julius Sipilä, Orion’s Discovery Data Technology Lead.
“Very few experts are capable of applying machine learning and AI methods to problems in chemistry, biology and pharmaceutical research. This is the expertise we’re seeking from universities.”
But how can AI support drug development? At the core of the process is the concept of a virtual laboratory, which the University of Helsinki and Aalto University have been developing in the FCAI Flagship. It is a digital research environment with accumulating data, based on which AI learns and can issue recommendations.
“We’re looking to transform research work in a pretty big way,” says Sipilä.
He notes that joining forces with the FCAI researchers felt natural thanks to their virtual laboratory idea and world-leading expertise. What Orion is seeking from its academic partners are data science knowledge and the ability to apply machine learning. The company is keen to drive development.
“Rather than standing on the sidelines, we wish to start using new methods as soon as they become available.”
Attractive opportunities for software companies
Since the project began in autumn 2023, Klami and his colleagues have identified the stages of early drug development where speed and precision could be increased. They are now developing new AI methods, which will eventually be tested with Orion. Among the solutions developed is an algorithm for optimising the conditions of a chemical reaction.
“We know to which of Orion’s core operations the virtual laboratory concept is suited,” notes Klami.
After the Co-Research project, the virtual laboratory will be commodified. Playing a key role at this point will be software companies, for which the project offers new business opportunities. One of the collaboration partners already involved is the software company Silo AI; others interested in the results include the pharmaceutical companies Admescope and Pharmatest.
A virtual laboratory can revolutionise research more broadly
If drug development becomes more efficient, it will be easier, for example, to bring innovative cancer treatments to the market and personalise drugs for patients. Sipilä believes the new ecosystem will also boost Finland’s security of supply by ensuring the country continues to have a viable pharmaceutical industry. He is pleased that research-driven projects helping to translate scholarly results into products can be completed with support from Business Finland.
“It’s great we have a financial instrument such as Co-Research.”
And that’s not all: the virtual laboratory can transform daily practices in several other fields too. Klami believes that, with small adjustments, the laboratory can be used in many different types of product development and research.
“Our aim is to ensure that AI methods need not be developed from scratch in each field.”
The collaboration in a nutshell
The project Virtual Laboratories for Pharmaceutical R&D has received Business Finland’s Co-Research funding for 2023 to 2025. The University of Helsinki and Aalto University have teamed up with the pharmaceutical company Orion to seek AI methods accelerating early-stage drug development. Collaboration partners include the software company Silo AI, the pharmaceutical companies Admescope and Pharmatest as well as CSC. Based on the project, the purpose is to build an innovative virtual laboratory supporting the creation of a new pharmaceutical ecosystem in Finland. The results can be used to introduce AI methods in other fields as well.
Further information about the collaboration
Arto Klami, Associate Professor, University of Helsinki Department of Computer Science
Julius Sipilä, Discovery Data Technology Lead, Orion Oyj
Text: Anna Humalamäki
Originally published on the University of Helsinki website.