Chemistry Nobel Given for Protein Discoveries

"The Nobel Prize in chemistry was awarded to David Baker, John Jumper and Demis Hassabis for their work to crack the code of the building blocks of life: proteins. The technology has the potential to transform drug development and the ability to understand human biology.

Jumper and Hassabis, both at Google DeepMind in London, designed the artificial-intelligence platform AlphaFold, which can accurately predict a protein's structure in minutes. Because of that, researchers can now better understand antibiotic resistance and create images of enzymes that can decompose plastic, the Nobel Committee said.

Baker, a biochemist at the University of Washington in Seattle, developed computational tools that enable scientists to design new proteins from scratch, with novel shapes and functions. His team has produced proteins that can be used in pharmaceuticals, vaccines, nanomaterials and tiny sensors.

"Proteins are the molecules that enable life," said Heiner Linke, chair of the Nobel Committee for Chemistry. "To understand how life works, we first need to understand the shape of proteins."

Wednesday's award is the second this week that recognized AI's ability to solve problems and transform whole industries. On Tuesday, the Nobel Prize in Physics was awarded to John Hopfield and Geoffrey Hinton for discoveries and inventions underpinning machine learning and AI.

On Tuesday, Hinton and co-laureate Hopfield reiterated concerns over AI's potential to harm humanity. Google DeepMind's Hassabis on Wednesday acknowledged that AI could be used for harm and that there should be international collaboration to mitigate those risks. But he also spoke to what he called its "extraordinary potential for good."

The breakthroughs from Hassabis, Jumper and Baker, buoyed by AI, have already led to changes in scientists' ability to understand existing proteins and create new ones.

Proteins are big, complex molecules that form bones, repair DNA and enable our immune response. They are made of a combination of 20 amino acids, strung together in endless combinations. Those strings fold and twist into shapes, and their structure determines their function.

If chemists know the sequence of amino acids in a protein, they should be able to predict the protein's structure. Yet the field struggled with that problem for over half a century. At one point, figuring out a single protein's structure could have been a student's entire Ph.D. work, said Mary K. Carroll, president of the American Chemical Society.

Hassabis -- a chess master and co-founder of AI company DeepMind -- and his team were able to help scientists to do just this. In 2018, his team developed a model called AlphaFold. It could predict protein structure with nearly 60% accuracy.

Jumper then joined the effort at Google, coming in with a background of theoretical physics and protein dynamics. The team also started harnessing neural networks called transformers, which can find patterns in large amounts of data in a more flexible way than before. In 2020, the team released AlphaFold2.

Baker, meanwhile, made it possible to build new proteins. The field of protein design had taken off, with researchers often tweaking existing proteins. Baker and his team wanted to create them from scratch, using a computer program called Rosetta to predict what amino-acid sequences put together could create the desired shape.

Rosetta could search a database of all known protein structures, look for snippets of similarities with the structure that was wanted, and then propose a string of amino acids that pieced them together. In 2003, Baker announced the creation of Top7, a protein with 93 amino acids that was entirely different to all known existing proteins. It was a first." [1]

1. U.S. News: Chemistry Nobel Given for Protein Discoveries. Abbott, Brianna.  Wall Street Journal, Eastern edition; New York, N.Y.. 10 Oct 2024: A.3.