Thank you. Listen to this article using the player above. ✖
Biotechnology and drug discovery are fields marked by constant innovation. Some emerging technologies of note have been bioengineering and, of course, artificial intelligence (AI) and machine learning. These tools are being leveraged to enhance scientific research and discovery in new and exciting ways, particularly by biotech start-ups.
At the European Laboratory Research and Innovation Group’s (ELRIG) Drug Discovery 2024 event, Europe’s largest drug discovery conference, Technology Networks spoke to a group of start-up companies beginning to make their mark in drug discovery. Featured in the Breakthrough Zone, this dedicated area on the show floor featured start-up biotech companies and provided them with the opportunity to showcase their innovative technologies and platforms.
We reached out to these companies to get the low-down on AI and bioengineering and their applications as emerging trends in biotechnology and drug discovery – here’s what they had to say.
Sarah Whelan (SW): How are AI and automation enhancing the productivity and precision of biotech/drug discovery, and what major trends do you think will shape this field going forward?
Luke Cox, CEO at Impulsonics: Any value in AI’s ability to synthesize huge amounts of information and transform it into actionable insights and lifesaving medicine will be entirely dependent on the quality of the data being fed in. I think the critical next step to unlock this is the automation of experiments to produce the high-quality datasets needed. Otherwise, we will simply be feeding these models noise and using a lot of energy to generate not a lot of useful information. Or to put it another way; garbage in, garbage out.
Jeroen Verheyen, co-founder and CEO at Semarion: AI and automation are transforming biotech and drug discovery by enhancing both productivity and precision by making research processes faster, more efficient and less prone to human error. Major trends in this space include AI-driven automation and a shift towards modular and flexible automation systems, which allow labs to adapt to new assays or shifts in strategy without the need for large, custom-built setups. This flexibility not only reduces infrastructure costs but also accelerates the response to emerging scientific challenges. However, many current systems still require robotics specialists, which slows adoption. In the future, more user-friendly platforms are expected to broaden access to these powerful tools.
Felix Lavoie-Perusse, co-founder and CCO at Saguaro Biosciences: The quality of datasets will become the most important variable in deriving meaningful insights from machine learning models in drug discovery. It’s been shown that bigger dataset sizes, better models and high-quality datasets drive better model performance.1,2
However, it is becoming clear that data and better-performing models are transforming into commodities. Indeed, there is an increasing number of public datasets being generated (i.e., image datasets from compound screens through the JUMP Cell Painting Consortium, or the Oasis Consortium initiative) and becoming freely accessible to be fed into AI models to output hopefully meaningful predictions.
In addition, we are seeing more AI models becoming freely available for everyone to use. Although models are costly to build and train, and only very well-funded companies like Meta can afford them, competitive market dynamics force these well-funded companies to make these costly models freely available to everyone. The defensive move by Meta to make their Llama 3 foundation model open source already gives a good indication of that trend.
This increased accessibility to data and AI models levels the playing field. This means that in order to differentiate and generate novel insights from AI models, organizations will need to generate unique datasets of high-quality. In the world of drug discovery, this means that organizations will need to use more physiologically relevant yet robust in vitro models. They will also need to use data-rich readouts and probes with limited impact on cellular biology, because the more relevant and information-dense the biological models are, the better the AI models will perform.
SW: What do you think are some of the most exciting developments currently in the field of synthetic biology/bioengineering, and how do you think these innovations will change the biotechnology/drug discovery landscape?
Helena Francis, chief of staff at Constructive Bio: Synthetic biology and bioengineering have huge potential to transform drug discovery and the entire bioeconomy. Particularly exciting is the ability to reimagine what is possible with biology, expanding the chemical space from which we can create new life-saving therapeutics and other biomaterials. Over the coming years, this will change the way we design and build biologic drugs, with new and improved chemical properties and more sustainable biomanufacturing at scale.
Ruizhi Wang, founder and CEO at Abselion: The recent advances in cell and gene therapy, built on many years of development in bioengineering, have demonstrated the power to impact human health in ways that we might never have thought possible. Pioneers have found a way to identify, engineer, produce and deliver therapies that can transform lives by offering potentially curative options.
Applying engineering principles to biological systems comes with many challenges associated with the inherent variability of a living cell factory. Key to accelerating access to next-generation therapies, and ultimately driving down costs, is improving consistency and scalability of viral vector production.
Optimizing the production process requires a multidisciplinary approach, and this is an exciting time with a shared momentum in the industry to tackle this head-on.