Nobel Prize-winning AI: AlphaFold’s breakthrough in protein structure prediction

The 2024 Nobel Prize in Chemistry has been awarded to David Baker, Demis Hassabis, and John Jumper for their pioneering contributions to computational protein design and protein structure prediction. This award highlights the profound influence of artificial intelligence (AI) on structural biology and its potential to reshape not only medicine and research but also industries like alternative proteins.

Hassabis and Jumper, both from DeepMind, received half of the prize for AlphaFold, an AI system that predicts protein structures with unprecedented accuracy. Knowing the shape of a protein is essential, as it determines the protein's function and interactions within living organisms. Previously, determining a protein structure required methods like X-ray crystallography or cryo-electron microscopy—both time-consuming and expensive. AlphaFold, on the other hand, can deliver these predictions within hours, transforming a process that could take years into a rapid, accessible resource.

Since its release in 2021, AlphaFold has made structural predictions for over 200 million proteins. These predictions, available for free in the AlphaFold Database, have democratized access to protein structure information, fueling innovation in numerous fields. Edith Heard, Director General of the European Molecular Biology Laboratory (EMBL), which collaborated with DeepMind on AlphaFold, praised its impact: “AlphaFold is the first AI system to send such ripples throughout the life sciences… unlocking new insights to address global challenges such as infectious disease, climate change, and food security.”

The implications of AlphaFold extend far beyond fundamental research. In the alternative protein industry, understanding protein structures is crucial for developing plant-based and cultivated meats that mimic animal products in texture, flavor, and nutrition. AlphaFold allows researchers and companies to model and design new proteins that are optimized for these qualities, supporting innovation in creating sustainable and appealing protein sources.

AlphaFold's open-access database is particularly beneficial for startups in the alternative protein space, enabling access to critical protein data without the high costs associated with traditional structural biology. Sergey Ovchinnikov, an evolutionary biologist from MIT, noted that the tool could significantly enhance food science: “With these tools, we're no longer limited to naturally occurring proteins.” This AI-driven protein design could accelerate the shift toward alternative proteins, contributing to a more sustainable global food system.

The other half of the Nobel Prize went to David Baker, a pioneer in computational protein design, whose work has paved the way for creating new proteins with tailored properties. At the University of Washington, Baker developed Rosetta and later RoseTTAFold, which not only predict protein structures but also enable the design of custom proteins. This capability has opened new pathways in synthetic biology, where engineered proteins can be used for a variety of applications, from medicine to environmental solutions.

Baker has built on his decades of research to enable the creation of proteins that could catalyze chemical reactions for sustainable energy, improve biomedical applications, or support agriculture. These innovations have reduced design times from weeks to seconds, empowering more researchers to engage in protein design. Reflecting on his journey, Baker said, “I have been really deeply inspired by others in the field and the people I’ve worked with… I stood on the shoulders of giants.”

Hassabis, who initially made headlines with AI models that mastered complex games like Go, views AlphaFold as the culmination of years of exploration into AI’s potential to tackle human challenges. “The long-lasting impact of AlphaFold will be defined by how researchers around the world use its predictions to gain new insights into how life works,” he explained. Jumper, a theoretical physicist who joined DeepMind with a fresh approach to protein dynamics, played a key role in refining AlphaFold's algorithms, which achieved levels of accuracy on par with experimental methods in 2020.

The AlphaFold Database has since catalyzed scientific discovery, inspiring thousands of published papers and spurring on related AI projects. EMBL-EBI Team Leader Sameer Velankar shared the optimism surrounding AlphaFold’s legacy: “The future is clearly bright for AI in molecular biology… leveraging AI and open access to large amounts of high-quality curated data [will lead to] many more research questions answered.”

"A few alt protein companies, such as Shiru, already use AlphaFold in their R&D process," revealed Noa Weiss, an AI & Machine Learning Senior Consultant. "But what has already been done is just a fraction of the potential AlphaFold has for the sector. Especially with the new and improved AlphaFold 3 released earlier this year, which is now capable of predicting not only the protein’s 3D structure but also its interactions with other molecules, the opportunities are vast. It could be used not only for the discovery of novel proteins, but also, for example, for the engineering of affordable media for cultivated meat, or to reduce the costs of precision fermentation. We are really only beginning scratch the surface of what it can do for our food systems."