Self-Driving Labs Accelerate Drug Discovery with AI and Automation

Automation and artificial intelligence are transforming how scientists design, test and refine novel molecules. At the University of Toronto, Stuart R. Green and the Acceleration Consortium are pioneering a self-driving lab poised to revolutionize the early stages of drug discovery.

The Rise of Autonomous Scientific Research

Drug discovery is traditionally a lengthy and expensive process, often taking over a decade and costing billions of dollars to bring a single therapeutic to market.¹ Autonomous laboratories offer a pathway to greater efficiency by utilizing robotic systems that can run experiments continuously, generate real-time data, and feed results into machine learning models. This new generation of labs integrates automated synthesis, biological screening, and computational intelligence into a seamless cycle of discovery.

Stuart Green and the Acceleration Consortium

Stuart R. Green, a Staff Scientist at the Acceleration Consortium (AC) at the University of Toronto, plays a pivotal role in developing these advancements.¹ His work focuses on developing automated biophysical assays and closed-loop workflows that connect chemical synthesis with biological evaluation within the Medicinal Chemistry Self-Driving Lab (SDL).¹

Overcoming the Hit-to-Lead Bottleneck

A significant challenge in drug discovery lies in the transition from identifying initial “hits” – potential drug candidates – to optimizing them into viable “leads.” Initial hits are often weak and lack selectivity.¹ Traditionally, medicinal chemists would synthesize and test hundreds of compounds to understand the structure-activity relationship, a process that can take months or years. The SDL’s approach aims to streamline this process.

Automated Synthesis and Data Analysis

The SDL constrains the search space to compounds that can be synthesized from a set of diverse building blocks using robust reactions.¹ It performs assays without compound purification in a direct-to-biology workflow on a fully autonomous system operating in a closed loop. This allows for the simultaneous synthesis and characterization of up to 100 compounds, significantly reducing the time between concept and result. Crucially, the system learns from its own output, using experimental data to train machine learning models that refine compound design.

Building an Integrated Ecosystem

Creating a self-driving lab requires more than just robots; it demands an integrated ecosystem where each component operates seamlessly. A key challenge was finding a chemistry-capable liquid handler that could perform chemical synthesis in an inert atmosphere.¹ The team developed a dual-handler setup – one for chemical synthesis and another for preparing biochemical assays – to meet performance and safety requirements.

The Benefits of Automation

While automation has long been used in biological screening, its application to chemical synthesis and assay integration is relatively recent. Automation enhances reproducibility, standardization, and data quality, which are crucial in early-stage discovery.¹ Automated systems are particularly valuable for screening small molecule libraries due to the repetitive nature of the task.

The Role of Artificial Intelligence

As AI becomes more integrated into laboratory workflows, the line between computation and experimentation blurs. AI can potentially suggest molecules that a human chemist might not consider, increasing the speed of candidate compound proposals and uncovering overlooked possibilities.¹ The long-term goal is a fully autonomous discovery engine where experimental data continuously refine predictive models.

Lessons for Labs New to Automation

For research groups exploring automation, Stuart advises leveraging the expanding ecosystem of commercial tools. He recommends discussing automation with individuals already working in automated labs.¹ When investing in equipment, teams should consider how instruments will be orchestrated, and whether to use commercial orchestration platforms or develop bespoke solutions.

The Future of Accelerated Discovery

Automation has the potential to broaden the scope of research by reducing the time and cost of hit-to-lead optimization, enabling scientists to pursue targets previously overlooked due to economic constraints.¹ This could facilitate progress towards treatments for rare diseases and conditions prevalent in low-income regions.

Stuart R. Green’s work demonstrates how automation is transforming early drug discovery. The self-driving lab is no longer a theoretical concept; it is already changing how new medicines are found.