New kids on the block, Atomic AI, recently closed a $35 million series A financing. DDW’s Diana Spencer speaks to Raphael Townshend PhD, Founder and CEO of Atomic AI, about the company’s successes so far, plans for the future and the unique approach of combining RNA structural biology and AI models.
DS: Congratulations on closing a $35 million series A financing. What advice do you have for other start-ups hoping to launch into this market?
RT: I believe the main reason for successfully closing our recent financing came down to the quality of the data we showed to our syndicate of investors. For founders planning to launch their start-ups, specifically drug developers, there needs to be sound, foundational science with promising data that will enable you to push further down the drug discovery pipeline. If you can get the life sciences community excited about the technology with compelling data, there is a high likelihood of obtaining financial support.
DS: The concepts behind Atomic AI’s platform are based on your own thesis on applying machine learning to the field of structural biology. How easy was it to transfer your studies into a viable business?
RT: Like many PhD-level scientists, my initial goal was to go the faculty route, launching my lab with grants and becoming a professor. However, the potential for my research to become a business was eminently clear once I began writing grants for what is now the technology behind Atomic AI.
First, there was a substantial commercial opportunity in building out Atomic AI’s Platform for AI-driven RNA Structure Exploration (PARSE). Using foundation models to discover 3D structural targets across various transcripts had massive potential in designing RNA-based medicines.
Second, we understood the need to scale up the engineering processes and data collection and increase the number of computational biologists and drug hunters to help validate the platform. These two factors together made it evident that a business was the best solution versus an academic research project. While I didn’t start out trying to start a company, the business direction made the most obvious sense for accomplishing our goals.
DS: How is Atomic AI “unlocking the next generation of RNA drug discovery”? What is unique about your engine? What have been the barriers to realising the full potential of RNA?
RT: Our goal is to build an engine with the ability to accurately and quickly model the 3D structure of RNA. A major barrier for the entire field is the limited RNA structural datasets that can be fed into AI models. However, Atomic AI has a unique structural engine that enables more intelligent and rational design of RNA-targeted and RNA-based medicines. This is what we mean by the “next generation of RNA drug discovery.” We can use our structural tools and AI-driven methodologies to create an entirely new way of identifying drug targets.
DS: How will tools that can accurately predict 3D RNA structures help us to drug the undruggable?
RT: A key challenge that has emerged is targeting a known RNA transcript associated with an “undruggable disease” with no knowledge of where to start on the transcript, more specifically, which parts of the transcript are even targetable through various therapeutic modalities. By identifying which parts are well structured, we are essentially pinpointing which transcripts are druggable, thus enabling us to selectively target those regions with functional molecules. This can potentially make “undruggable diseases” become “druggable.”
DS: Your engine leverages RNA structural biology and AI foundation models. What are the advantages of bringing these breakthrough technologies together?
RT: There has been tremendous progress in both fields over the last few years. For RNA, we’ve seen its success with Covid-19 vaccines and the approval of the first-ever RNA-targeting small molecule, risdiplam, for treating spinal muscular atrophy (SMA). On the AI side, a vast array of technologies has tapped into this powerful tool to create some of the most advanced large language models such as ChatGPT. Synergising RNA with AI enables further progress on the RNA drug discovery frontier. It is excellent timing to combine these technologies and the right time for Atomic AI.
DS: What are the benefits of exploiting AI/machine learning in this sector? What are the challenges/limitations?
RT: AI/ML can be incredibly powerful when used appropriately. It is essential to understand that AI/ML is not a quick fix that can solve every problem. Therefore, realising where AI/ML can make dramatic improvements is vital. A basic example would be when you have significant amounts of data you can exploit, an AI learning algorithm can do better than any hand-designed, manual computational system.
For the biotech sector, we can only expect these algorithms to be as good as the data fed into the technology. To an extent, we have the opportunity to get around this challenge by using foundation models where we can generalise relatively easily through training core algorithms, even with limited data. For Atomic AI, we can potentially achieve generalisation by understanding and exploiting the underlying structure of RNA molecules. This means we can train the physics-informed AI algorithms on a limited set of RNA then use them to analyse completely novel RNA transcripts.
DS: AI is already playing an important role in drug discovery. Do you think there is untapped potential there?
RT: Definitely yes. We’re still scratching the surface. Much of the high-level AI research has not been focused on drug discovery or the biotech space, mainly because these areas have been inaccessible to the general AI/ML community. By bridging the gap and bringing world-leading AI researchers into the life sciences, we have the opportunity to create safer and more effective therapeutics for a broad range of diseases.
DS: You are planning to use the engine yourselves rather than sell it to others. Do you have any therapeutic targets in mind for your first small molecule drug candidate? Is there anything currently in the pipeline?
RT: Currently, we are developing our internal pipeline. While we are not disclosing specific targets at the moment, we believe there is enormous potential for RNA-targeting small molecules in the therapeutic areas of oncology, neurodegeneration, neuromuscular disease, infectious disease, rare disease and immunology.
DS: You were recognised in the 2023 Forbes’ 30 under 30 for Science. To what do you attribute your success at such a young age?
RT: I believe it is imperative to be open to jumping into new areas and getting out of your comfort zone. I started my PhD work in computer vision but decided to hop into structural biology. With no guard rails, it can be challenging to learn new things. But the willingness to incessantly ask questions is critical. If you can get over that, you can get into a space where few, if any, other scientists have gone. I saw that the field of structural biology was fairly neglected by the AI space. By digging into the research and learning from RNA and AI experts, I knew there was a fantastic opportunity to merge the two fields.
DS: How do you think the RNA market will progress in the next five years and where will the opportunities be?
RT: We are just getting started in the RNA space. The recent wins with the Covid-19 vaccines and the risdiplam approval demonstrate the potential of RNA. Over the next five years, there will be incredible developments in the next generation of RNA-based medicines such as mRNA vaccines and circular RNA, and RNA-targeting medicines, like small molecules and antisense oligonucleotides. We will also see improved gene therapies and advanced CRISPR guide RNA design tools. It is exciting to see where the RNA market will go from here.
DS: What advice would you give to others looking to turn scientific knowledge into a profitable business?
RT: Make sure there is a business case behind what you are trying to accomplish. Even if, from an academic standpoint, the science you are pursuing might be groundbreaking, it is essential to realise that not everything scientifically interesting will lead to a viable business at the end of the day. Therefore, deciding which parts can potentially be commercialised is a key consideration before launching a business.
Also, I urge academic and scientific founders to tap into the knowledge from those advisors who have already laid the groundwork in the entrepreneurial space, like fellow biotech founders and potential investors, to ensure that your ideas are sound. As academics entering the business world, we often know next to nothing when we start. So being willing to put yourself out there is extremely important.
Prior to founding Atomic AI, Dr Raphael Townshend was a PhD fellow at Stanford University, where he wrote his thesis on Geometric Learning of Biomolecular Structure and taught in Stanford’s machine learning and computational biology programs. During his PhD program, Dr Townshend also held positions at DeepMind and Google on their artificial intelligence and software engineering teams, and he founded the inaugural workshop on machine learning and structure biology. Dr Townshend holds a BS in Electrical Engineering and Computer Science from the University of California, Berkeley and a PhD in Computer Science from Stanford University.