And in the end, artificial intelligence wins. After rewarding the theoretical bases behind neural networks, the fundamental basis of modern AI, with the Nobel Prize in Physics on Tuesday October 8, this Wednesday, the Swedish Academy is rewarding one of the applications of these neural networks: Alphafold, an AI designed by Google DeepMind. John Jumper and Demis Hassabis, director and co-founder of Google DeepMind respectively, have developed an AI capable of predicting the three-dimensional structure of proteins. They share their 2024 Nobel Prize in Chemistry with David Baker, an American scientist who has worked for more than thirty years on this question of protein shape. He developed a method to create an artificial protein.
“I’m not surprised that they are rewarded but I am surprised that it happens so soon, Alpha Fold is only four years old”comment for Release Sophie Sacquin-Mora, CNRS research director at the Theoretical Biochemistry Laboratory. In fact, John Jumper, 39 years old, Demis Hassabis, 48 years old and David Baker, 62 years old, drastically reduce the average age of the 2024 winners which until now stood at over 70 years old. The Nobel Prize usually takes decades to be sure of rewarding a crucial advance in a sector.
Sophie Sacquin-Mora remembers «l’effervescence» when Alpha Fold was released in 2020. “Many teams adopted the software, there were very quickly hundreds of articles using it”she testifies. And for good reason, understanding the three-dimensional conformation of a protein is essential in biology but until now it was a tedious task. The composition of proteins is easily known. These molecules acting in the body are created by genes. When we know the sequence of a gene, we know the sequence of amino acids that make up the associated protein. But the form that the latter takes remains a mystery. Scientists have several imaging techniques, but it takes months or even years to get an answer. Result, “we know the structure of 200,000 proteins out of the millions that exist in nature”believes Sophie Sacquin-Mora. There is therefore still work to be done and AlphaFold will speed it up since it allows us to have a proposed structure much more quickly, even if it is not perfect.
“Life’s work”
“AI was able to tackle the subject because the structural biology community decided very early on to pool its data”points out Konrad Hinsen, researcher at the Center for Molecular Biophysics. In fact, the most powerful AI is nothing without a quality database to train on. This is where the award awarded to David Baker, involved in structural biology for thirty years, comes into play. “This award recognizes a lifetime’s work culminating in protein engineering”believes Konrad Hinsen. David Baker developed a computer program to create the ad hoc protein depending on the function you want it to perform. For example, specifies the Nobel committee, this technique can be used to produce new drugs, vaccines, or tiny sensors. Basically, the program compares the desired function to the functions of known proteins and proposes the sequence of amino acids which should give rise to the protein capable of fulfilling the requested function.
In 2023, the Swedish Academy distinguished three researchers based in the United States, Moungi Bawendi, from the Massachusetts Institute of Technology (MIT), Louis Brus from Columbia University, and Alexei Ekimov who works for Nanocrystals Technology, for “the discovery and synthesis of quantum dots”, major advances in the field of nanotechnology. The 2024 Nobel season was launched on Monday with the medicine prize, awarded to Americans Victor Ambros and Gary Ruvkun for their discovery of microRNA. It continues with literature on Thursday, the Nobel Peace Prize on Friday, and that of economics on October 14.
