Summary

• Self-driving labs (SDLs) can speed up scientific discovery and technological progress, supporting growth and innovation in the UK. 
• A new UK government should invest £60 million in self-driving labs, and automated science, expanding on current investments such as the £12 million investment into the AI for Chemistry research hub at the University of Liverpool.
• SDLs should focus on neglected but high-leverage domain applications, open-source modular software and hardware components. 
• Public research organisations should act as data collection centres to streamline data collection, processing, and management for SDLs. An existing example is Imperial College London’s White City Campus for microfabrication research.
• The government should also support SDL-related workforce training by funding traineeships, fellowships, and career development grants and integrate automation engineering into existing academic programmes.

Challenge and Opportunity

The UK has a productivity problem. Single US universities and industry firms outpace the country in research areas shaping the future, like synthetic biology, quantum, and AI. We should be more proactive and strategic in leveraging our talent pool and top-tier research institutions to generate breakthroughs. One approach is to invest in automating and innovating the scientific process itself through self-driving labs (SDLs).

Self-driving labs (SDLs) are automated platforms and robots capable of independently executing tasks within a laboratory setting, particularly repetitive ones. They conduct experiments rapidly, precisely, and consistently, reducing the likelihood of costly errors. With the data they generate, SDLs accelerate data-driven discovery and complement scientists, enabling them to concentrate on high-impact, creative scientific work.

Currently, the development of a new drug takes around 10 to 15 years and costs approximately £1.15 billion from ideation to distribution. With SDLs, this timeframe could be vastly reduced, particularly in the hit-to-lead phase (lasting approximately six months) and lead optimization (taking around two years). 

Similarly, the journey from concept to market for materials takes about 20 to 25 years. Automation could help materials science discovery go from a few decades to a few years and bring it down to a 100th of the cost.

Software is revolutionising the scientific landscape, with industry labs leading the charge in its development. In the field of life sciences, understanding protein structure is crucial for drug design. Traditionally, predicting protein structure has been an arduous, time-consuming, and expensive process. However, AlphaFold, an AI-powered protein structure predictor developed by DeepMind, has transformed this task, making it almost instantaneous. 

Similarly, DeepMind has also created GnOME, an AI system that automates the discovery of novel materials. These advancements demonstrate how software and AI are reshaping scientific research, enabling faster and more cost-effective discoveries across various domains.

Others are already investing in SDLs. For example, the University of Toronto’s Acceleration Consortium is leading efforts for SDLs in materials science. The consortium was funded with £160 million ($200 million USD) in 2022 by the Canada First Research Excellence Fund (CFREF).

SDLs harness the power of cutting-edge software, advanced hardware, and massive datasets to supercharge scientific discovery. The UK has the opportunity to lead the world in this technological revolution, transforming the way we conduct research and accelerating the pace of innovation.

The Precedent for Automated Science in the UK

While SDLs represent a new age for scientific research in the UK, the country already has a track record of exciting progress in leveraging automation for scientific research:

• In 2009, The University of Liverpool engineered the first robot scientist— Adam, with funding from the Biotechnology and Biological Sciences Research Council (BBSRC). Adam autonomously conducted experiments optimising hydrogen production and identifying genetic processes in yeast. 
• Eve, a robot developing treatment for malaria, was developed at the King Lab, now at the University of Cambridge.
• The University of Liverpool and Imperial College London are leading the UK's £12M AI for Chemistry Hub, supported by the Engineering & Physical Sciences Research Council (EPSRC) in 2024.
• In early 2023, Automata partnered with the Advanced Sequencing Facility (ASF) at The Francis Crick Institute to automate genomic sample preparation.

These examples show how historically the UK’s researchers and institutions have successfully worked on automated scientific discovery. By investing in SDLs, the government can build on this strong foundation, leveraging the country's expertise and infrastructure to drive scientific breakthroughs and maintain its position at the forefront of innovation.

Plan Of Action

1. Invest in Open-Source and Modular Components

The Department for Science, Innovation, and Technology (DSIT) should allocate £40 million in grants through the UK Research and Innovation (UKRI) and an open call for proposals for open-source and modularised software and hardware and technical safeguards.

This investment, totalling £40 million, will primarily fund the development of open-source and modularized software (£10 million) and hardware components (£30 million) for SDLs. Additionally, it will support complementary initiatives, such as research and development for technical safeguards against biological risks associated with SDLs and enhancements to data infrastructure.

Modularity, or functional compartmentalisation, is key to the success of SDLs. By designing and manufacturing automated components separately (e.g., robotic arms and microplate readers), researchers can combine these components to execute tasks and adjust them to different complex workflows. 

These modular components should automate a wide range of useful, repetitive, and standardised lab functions, such as:

• Assays (determining the quality and characteristics of a substance), using microfluidics or "lab-on-a-chip" (LOC) technology and genetic material isolation with minipreps.
• Dispensing and mixing chemicals, managing conditions like temperature and pressure during chemical reactions in experiments.

These innovations can streamline investments into specific areas of R&D, including:

• Cloud-based computing systems, sensor arrays, communication networks, mapping technologies, testing facilities, and simulation environments for training.
• Foundational hardware components tailored for SDLs, like microfluidics, sensors, and circuits.
• Data storage, preprocessing, and management techniques, such as distributed storage systems or secure data-sharing protocols.
• Security and privacy of SDLs, including encryption techniques, secure computation protocols, and identifying high-risk cases for data collection and synthesis.

To ensure data security and compliance, the UKRI and the SDLs programme should work with the Data and Analytics Research Environments programme (DARE UK) to implement federated architectures for large-scale SDL programmes. 

DARE UK leads the DARE UK Federation, which focuses on Trusted Research Environments through federated architecture and works on the UK's national digital research infrastructure. With federated architecture, data stays local instead of going to a centralised web server. The hope is that sensitive data can be regulated but secure, keeping potentially sensitive biological data safe.

By consulting with DARE UK, the SDLs programme can develop SDL-specific digital infrastructure priorities, create funding calls, and pilot federated architectures for data collection centres established at large laboratories. 

Furthermore, these organisations can enhance biosecurity measures for sensitive data related to SDLs, thereby reducing potential risks associated with datasets that could be misused to engineer viral pathogens or develop biological agents capable of compromising human immune systems.

2. Fund Major Research Centres

Through UKRI, DSIT should directly fund £50 million in 2-5 major research centres as a pilot to establish SDL programmes tailored to their specific needs, investing in components to meet specific institutional needs. 

Eligible research centres would include those registered as charities in the UK, such as university laboratories or independent institutes registered as charities in the UK.

SDLs at large laboratories can help speed up progress for research programmes that already have momentum. The data collected from these programmes can help the UKRI scale SDL programmes. Large laboratories have established track records that make it easier to calculate SDL productivity gains, understand which components are the highest-leverage to manufacture, and survey scientists’ opinions on SDLs.

By focusing on modular components and housing them at large labs, many scientists can take advantage of them and benefit from having a shared research agenda. Localised SDL facilities at large labs may even encourage interdisciplinary collaboration and, as a result, more innovation.

By funding automation programmes at such facilities, the initial investment can be spread across a larger number of tasks and researchers. 

Examples of programmes to fund include the proposed Embedded Automation Science Technology Platform (STP) at the Francis Crick Institute, which employs machinery to perform common, time-saving tasks across their laboratories. Other potential laboratories to fund include the Quadram Institute and the John Innes Institute, both on UKRI's list of research institutes eligible for funding.

3.  Establish Data Collection Centres

The DSIT should require suitable public research organisations using SDLs to become data collection centres. These centres can efficiently collect, manage, and analyse the vast amounts of data generated by the automated scientific discovery process facilitated by SDLs. The data produced by these centres can contribute to large-scale (public) scientific databases like the UK Biobank and the NIST (National Institute for Standards and Technology) Materials Data Repository.

The Biofab at Imperial College London's White City Campus, which uses microfabrication technology for biomedicine, is an example of a programme with the potential to become a data collection centre. Biofabrication is microfabrication applied to biology (and medicine). It is a process that involves the use of biological materials, cells, and/or biomaterials to create functional structures or devices. Generated data that could be of public scientific use include gene expression data and biomaterial properties.

FAQs

How can we train researchers to use SDLs?

Researchers can be trained to use SDLs through:

• Traineeships, Fellowships, and Visiting Programmes: Offering opportunities for hands-on experience and specialised training in SDL technologies.
• Career Development Grants: Providing financial support for researchers to enhance their skills in automation engineering relevant to SDLs.
• Incorporating Automation Engineering into Academic Programmes: Integrating modules or courses on automation engineering into existing academic programmes to equip researchers with the necessary skills.
• Mandating SDL Exposure in Training Programmes: Requiring exposure to SDLs as part of mandatory training programmes for laboratory personnel to ensure broad familiarity with SDL technologies.
• Establishing Specialized Arms under EPSRC Centres for Doctoral Training: Creating dedicated arms within existing doctoral training centres, collaborating with relevant thematic areas to ensure comprehensive training in SDL-related skills.

By supporting these initiatives, we can effectively train researchers to interpret and utilise data generated by SDLs, integrate SDLs into their workflows, and understand the associated risks and ethical considerations.

Will SDLs replace scientists? How will they alter the research job market?

SDLs automate routine tasks for scientists, freeing up time for data analysis and theory validation. However, they may favour larger labs with tech support. Open-source software could level the playing field for smaller labs.

What are the potential applications of SDLs in specific areas of science?

SDLs can help synthesise and test new materials, engineer tissues with microfluidic chips with organ-on-a-chip technologies, and characterise genomes. The automation provided by robotic arms enhances efficiency across diverse areas, enabling tasks such as sample handling, pipetting, plate manipulation, high-throughput screening, and precise measurements.

What are the biosecurity implications of SDLs?

SDLs have the potential to streamline repetitive procedures and accelerate the design-build-test-learn (DBTL) cycle in biological research. However, this rapid experimentation process could also expedite the development of harmful biological agents, raising biosecurity concerns.

While this proposal focuses on SDLs for automation that carry out specific, arranged tasks, more intelligent SDL systems using AI could introduce additional risks at multiple stages of the DBTL cycle:

1. In the design and learning phase, AI-enhanced SDLs could:
2. 1. Propose new harmful biological agents
   2. Improve future DBTL cycles by interpreting experimental results
   3. Use AI algorithms to optimise pathogen bioreactors
   4. Help target location selection
2. In the learning and testing phases, AI-enhanced SDLs could:
3. 1. Quickly and autonomously design biological agents with desired properties by iterating on model predictions and improving the design process
   2. Create delivery mechanisms to optimise infectious doses and ensure environmental survival in given delivery vehicles

To mitigate these risks, UKRI funding applications already require applicants to outline their data management and sharing practices, consider ethical implications, implement responsible research innovation (RRI) best practices, and plan for genetic and biological risk mitigation. R&D investment from DSIT, facilitated through the UKRI, should cover various aspects of SDL research and infrastructure, including the development of technical safeguards.

Like all applicants to relevant UKRI funding schemes, those working on SDL-related projects should adhere to the Medical Research Council (MRC) guidance for using the Joint Electronic Submission (Je-S) system or UKRI Funding Service. Ultimately, recipients of public research funding for SDL projects must effectively anticipate their projects' risks and outline strategies to mitigate them.