AstraZeneca iLab: The automated lab of the future

Summary: Artificial intelligence (AI), automation and robotics have the potential to transform how we discover and develop new medicines. As we embed these capabilities across our R&D, they offer the prospect of making better decisions faster. The AstraZeneca iLab in Gothenburg, Sweden, is our prototype of a fully automated medicinal chemistry laboratory which, through seamless integration with our Molecular AI group, is propelling the design-make-test-analyse cycle (DMTA) of drug discovery to new heights of efficiency.

Speeding up drug discovery to get medicines to patients

The concept of automation in drug discovery is not new but the complete automation of the DMTA cycle has not yet been achieved and is the ambition of the AstraZeneca iLab. The potential of automation to transform medicinal chemistry is huge, especially when integrated with AI and machine learning capabilities.

Small molecule drug discovery is driven through multiple iterations of the DMTA cycle. This includes designing entirely new molecules, making them through chemical synthesis, testing them in a series of biological assays and analysing any improvements made, before starting on the next round of design. It is a long and time consuming process. AI, automation and robotics have the potential to drive this cycle much more rapidly and our aim is ultimately to identify potential drug candidates in half the time it takes today.

We started this journey in 2017, with an ambition to build and optimise a prototype that could automatically synthesise small molecule compounds, purify them and make screening-ready solutions for testing. Once the compounds have been tested, AI steps in to analyse the data and suggest new compounds to make and test.

We have also developed a new make-and-test technology called nanoSAR, a miniaturised high frequency synthetic process coupled with biophysical screening, which is allowing us to explore a wide range of molecules around a key lead compound much more quickly.

How AI and machine-learning are transforming what to make next and how to make it

The iLab works closely with our world-leading Molecular AI group which drives the ‘design’ and ‘analyse’ elements of the DMTA cycle – in other words ‘what to make next’ and ‘how to make it’. This group harnesses AI and machine learning to help our chemists make better decisions faster. A recent paper from the team published in Nature Machine Intelligence describes novel AI-based models that use conditional recurrent neural networks to enable our chemists to work interactively with computers to speed up the exploration of chemical space and the design of potential new drug molecules.

Holistically uniting chemistry, automation and artificial intelligence technology

We are still on the journey but have achieved so much over the past two to three years. We are now capable of synthesising several small molecule compounds in parallel and automatically purifying them. Much of the traditionally manual laboratory tasks can now be done by robots. Working with specialist vendors in hardware and software, primarily BioSero and Zinsser Analytic (now part of Ingersoll Rand), we have reached the third generation of our prototype platform, which is improving every day. And in the next few years, we aim to have a fully functional automated chemistry lab both in Gothenburg and in Cambridge, UK.

The iLab has also been featured in an article on innovations in automating drug discovery in Nature Reviews and showcased in a recent meeting report in Nature Chemistry.