[Credit: Cellarity]
The traditional drug discovery approach is broken, but so too are the approaches of many AI-focused organizations seeking to reboot the process. That’s the perspective of Fabrice Chouraqui, CEO of Cellarity and CEO Partner at the investment firm Flagship Pioneering, which played a role in launching Moderna in 2010. While a growing number of companies are focused on using AI to streamline drug discovery, the approach is still something like gambling with long odds. Traditional drug developers tend to “place a bet on a single molecular target very early on,” he said.
Despite the dizzying pace of scientific development, the fundamental approach of drug discovery has seen limited evolution. While organizations are exploring strategies to redefine the process, they often bring a new tool to an existing process. With new technologies like AI, the initial thought is often to apply it to familiar territory. They take a look at their current process, and explore how AI could fit into each step in the process. “But they haven’t thought about a complete overhaul of the process,” Chouraqui added. What’s needed is a shift in perspective on how AI and other technologies can change the game. “Ultimately, you don’t want a drug to only hit one target. It might work for some diseases, but it’s too reductionistic,” he said.
Cellarity is aiming for a complete overhaul
Fabrice Chouraqui [Credit: Flagship Pioneering]
While the traditional target-based drug discovery process is well-validated, it remains reductionist and has “barely evolved over the years,” Chouraqui observed. As technologies like AI advance, many organizations instinctively try to retrofit these advancements into their existing frameworks. They evaluate their current procedures, seeking to insert AI wherever possible. “But they haven’t thought about a complete overhaul of the process,” Chouraqui added.
Cellarity aims for such a shift in perspective. Instead of focusing on a single molecular target, the company is taking a cellular approach. By focusing on the entire cell — a complex, interconnected system — it aims to unravel the complex web of disease biology. “It is a totally new way of doing drug discovery,” Chouraqui said.
The process is non-trivial, he acknowledged. “Cells are made up of billions of interacting molecules,” he said. “This is where multi-omics data and AI come into play.” Omics technology makes it possible to translate cellular biology into data points and, consequently, end up with a much higher-resolution understanding of biology. “By shifting our focus from a single molecular target to the underlying cellular behavior, we can unravel the complexity of disease biology and develop medicines that are out of reach with the current drug discovery process,” Chouraqui said.
The power of ‘what if?’
Flagship Pioneering, which founded Cellarity in late 2019, thrives on challenging conventional wisdom. “Everything starts with a ‘what if’ question,” Chouraqui said. Moderna’s origins, for instance trace back to the question, “What if we could turn our own cells into personal protein factories?” That led to mRNA. We start with big questions that could be very disruptive, and then we try to either find or develop the science that goes with it. In the case of Cellarity, the initial question was: “What if we could design drugs that could target the whole cell instead of just a single molecular target?”
Cultural dimensions of AI-driven drug discovery
It goes without saying that a novel approach to drug discovery requires expertise and interdisciplinary collaboration. “Talent is the name of the game,” Chouraqui said. “We’re nurturing a new generation of talent who are, in a sense, multilingual.” Because the company operates at the confluence of diverse fields such as biology, chemistry, multiomics, and computation, interdisciplinary communication and collaboration are paramount. Such teamwork is “at the core of this new era in scientific research,” Chouraqui noted.
While advances in computation and algorithms are providing new insights into biology, offering significant scientific possibilities, “AI is just part of the picture,” Chouraqui said. The effectiveness of machine learning hinges on the underlying algorithms, which in turn are only as reliable as the data they are built on. “So you need to be strategic and understand how you can harness those new technologies and the new computational power to benefit science and patients,” he added.
It’s not good enough to harness AI to speed up traditional drug development — if the resulting drugs aren’t better than the standard of care. “If you develop a drug with a different mode of action, everyone is happy about it,” Chouraqui said. “But if it has the same safety and efficacy profile as treatments used for the past decade, then its value is questionable.” Payers or doctors would have little incentive to back a new drug with no discernible advantages.
Cool science isn’t enough
In an era where the rate of scientific advancements in biotech and drug discovery is unparalleled, it’s tempting to get caught up in the allure of novelty. New drugs need to be differentiated — not just in terms of its mode of action but also in terms of efficacy and safety.
If anything, the surging interest in AI underscores this reality. “It’s challenging, especially when working with fascinating technologies – where the possibilities are immense,” Chouraqui said. “You can lose yourself. And you can lose sight of the importance of value creation.”
Yet, amidst the allure of technological wonders and the pursuit of the next big discovery, priorities should remain clear. “Simply doing cool science isn’t enough. It has to be cool science that can lead to something differentiated — creating value for patients,” Chouraqui concluded.