In the heart of South London, nestled within the walls of a repurposed biscuit factory, a revolution is brewing. Where once stood giant mixers and industrial ovens, now reside robotic arms, incubators, and DNA sequencing machines. This is the home of LabGenius, a company led by James Field, which is pioneering an AI-driven method to engineer medical antibodies. Antibodies, nature’s frontline soldiers in our immune system, are protein strands designed to latch onto foreign invaders, aiding their expulsion from the body. Since the 1980s, synthetic antibodies have been produced by pharmaceutical companies to combat diseases like cancer and reduce the risk of organ transplant rejection. However, the design process for these antibodies is laborious and time-consuming, requiring protein designers to sift through millions of potential amino acid combinations.
Field, founder and CEO of LabGenius, saw an opportunity to streamline this process. In 2012, while pursuing a PhD in synthetic biology at Imperial College London, he noticed the declining costs of DNA sequencing, computation, and robotics. He envisioned a way to automate the antibody discovery process using these technologies. LabGenius’ approach is a blend of machine learning and automation. In their Bermondsey lab, algorithms design antibodies to target specific diseases. Robotic systems then construct and cultivate these antibodies, run tests, and feed the data back into the algorithm, all under minimal human supervision.
The process begins with human scientists identifying a range of potential antibodies for a specific disease. They need proteins that can distinguish between healthy and diseased cells, adhere to the diseased cells, and then summon an immune cell to eliminate them. LabGenius’ machine learning model can navigate this vast search space much more efficiently. The model selects over 700 initial options from a pool of 100,000 potential antibodies. It then designs, builds, and tests them, aiming to identify promising areas for further investigation. The testing process is almost entirely automated, with high-end equipment preparing samples and running them through various stages of testing. The results from the first batch of 700 molecules are then fed back into the model, refining its understanding of the space.
As the algorithm learns more about how different antibody designs affect treatment effectiveness, it becomes more adept at balancing the exploration of new areas with the exploitation of promising designs. Conventional protein engineering often involves making small tweaks to a molecule that shows some promise. However, this can lead to a focus on optimizing a suboptimal solution, while potentially more effective options are overlooked. LabGenius’ approach, on the other hand, can uncover unexpected solutions more quickly, taking just six weeks from problem setup to the completion of the first batch. LabGenius has raised $28 million from investors like Atomico and Kindred and is starting to collaborate with pharmaceutical companies, offering its services in a consultancy capacity. Field believes their automated approach could be applied to other forms of drug discovery, transforming the traditionally lengthy and artisanal process into something more efficient.
Ultimately, Field envisions this leading to better patient care, with antibody treatments that are more effective and have fewer side effects than those currently designed by humans. “You find molecules that you would never have found using conventional methods,” he says. “They’re very distinct and often counterintuitive to designs that you as a human would come up with—which should enable us to find molecules with better properties, which ultimately translates into better outcomes for patients.”
