Chemistry Meets Code: AI’s Promise in Drug Discovery

Discussion with Deren Koseoglu

In an industry marked by complexity and rapid technological advancement, Deren Koseoglu, VP of Global Accounts at eMolecules, is at the forefront of transforming drug discovery. With a background in chemistry and extensive experience in commercial roles, Koseoglu emphasizes the potential of computational chemistry and artificial intelligence (AI) to revolutionize the pharmaceutical landscape.

The State of Drug Discovery

Despite technological advances, drug discovery remains hindered by inefficiencies in R&D. Koseoglu describes the industry as “bloated with research and development (R&D) inefficiencies”, highlighting how the traditional approach is slow and costly. “The industry continues to follow ‘Eroom’s Law,’ the inverse of Moore’s Law, where drug discovery is becoming progressively slower and more expensive over time”, he explains, referring to how innovation costs are rising without a corresponding increase in output. However, the integration of AI and machine learning offers hope, especially for streamlining the preclinical phase of drug development. These computational techniques promise to speed up processes and reduce costs significantly, making them crucial for future advancements.

Koseoglu notes that despite the potential for computational techniques to accelerate R&D, many pharmaceutical companies are still entrenched in traditional methods. This reluctance is partly due to the massive investment and infrastructure built around established R&D processes. However, the pressure to innovate and reduce costs is driving a gradual shift towards embracing new technologies. “Pharma has to adapt or risk being left behind”, Koseoglu asserts. He points to a few pioneering companies that have already made significant progress by integrating AI and machine learning into their workflows, showcasing the potential of these technologies to transform drug discovery.

Investment Opportunities in Pharma

For investors seeking lucrative opportunities in the pharmaceutical sector, Koseoglu recommends focusing on companies with a robust computational presence. He suggests that “investing in preclinical companies offers high returns if you pick a winner”, as AI-driven drug discovery firms hold immense potential for innovation. The rise of tools like AlphaFold, which accurately predicts protein structures, showcases the transformative power of computational methods. By investing in these forward-thinking companies, investors can capitalize on breakthroughs that promise to reshape the drug development landscape.

Koseoglu emphasizes the importance of identifying companies that are not only adopting computational techniques but are also leaders in integrating these methods into their R&D processes. “The real value lies in companies that use AI to complement and enhance traditional drug discovery, not just replace it”, he advises. For investors, the challenge is to distinguish between companies genuinely leveraging AI to drive innovation and those merely using it as a buzzword to attract funding. Koseoglu suggests looking for firms with a proven track record of successful projects and partnerships in the computational space as key indicators of potential success.

The Role of AI in Drug Discovery

AI and machine learning (ML) are revolutionizing how small molecules are discovered, addressing the vast chemical spaces that researchers must navigate. Koseoglu explains, “With computational tools, we can screen billions of compounds, drastically cutting down development time”. This approach was exemplified by Schrödinger in a recent success story, where they discovered a development candidate targeting MALT1 in just ten months using AI and physics-based methods. By computationally screening over 8 billion compounds and synthesizing only 78, they significantly accelerated the process before nominating their development candidate. Such advancements demonstrate the potential of AI and physics-based methods to not only expedite drug discovery but also enhance accuracy and reduce costs.

One of the key advantages of using AI in drug discovery is its ability to predict the properties of molecules before they are synthesized, thus reducing the need for extensive laboratory testing. Koseoglu describes this as a “game-changer”, allowing researchers to focus their efforts on the most promising candidates and avoid costly dead ends. Additionally, AI can identify patterns and relationships within vast datasets that would be impossible for humans to discern, leading to new insights and potential breakthroughs. “AI doesn’t just speed up the process; it opens up entirely new avenues of research that were previously inaccessible”, Koseoglu notes.

Collaboration vs. Intellectual Property

Balancing collaboration with intellectual property protection is a crucial aspect of modern drug discovery. “Collaboration is key”, Koseoglu stresses, “but balancing it with IP protection is crucial”. He advocates for industry-wide cooperation, particularly in developing computational techniques, to push scientific boundaries and improve patient outcomes. Despite the challenges posed by political and regulatory environments, fostering collaboration could lead to significant advancements. By working together, organizations can enhance research efforts and drive innovation in drug discovery.

Koseoglu acknowledges the inherent tension between the need to protect intellectual property and the benefits of collaboration. He suggests that the industry could benefit from more open-source initiatives and shared databases that allow researchers to build on each other’s work. “By sharing data and techniques, we can accelerate progress across the board”, he argues. However, he also recognizes that this requires a cultural shift in an industry traditionally focused on secrecy and competition. “It’s about finding a balance that encourages innovation while still protecting the commercial interests of those who develop new therapies”, he concludes.

Looking Ahead: The Future of Drug Discovery

As the industry evolves, Koseoglu remains optimistic about the role of AI in drug discovery. He highlights the importance of generating more data to feed these models, emphasizing the potential of improved computational power. While some predict an AI bubble burst, Koseoglu envisions a “slow deflation to reality”, where validated data and successful companies thrive. By embracing AI and computational methods, the pharmaceutical industry can navigate future challenges and continue to innovate, ultimately benefiting patients and healthcare worldwide.

Looking to the future, Koseoglu foresees AI playing an even more integral role in the drug discovery process. He predicts that advances in quantum computing and other technologies will further enhance the capabilities of AI, enabling researchers to tackle even more complex problems. “The potential for AI to revolutionize healthcare is immense”, he says, “but we must continue to invest in research and development to fully realize its benefits”. As AI becomes more sophisticated, it will likely drive new discoveries, improve patient outcomes, and reduce healthcare costs, making it an essential tool for the pharmaceutical industry.