A person born after 1985 won the Nobel Prize and received top honor at such a young age. What did he rely on?

Recently, with the announcement of the 2024 Nobel Prizes, the scientific community has been discussing this event with unprecedented enthusiasm. In addition to the fact that the award went to AI twice, the most popular one is John Jumper, the first "post-85" winner in the history of the Nobel Prize. Many people wonder: How could he win the Nobel Prize at such a young age? What exactly did this genius study that helped him win the top honor that all scientific researchers in the world dream of? Let us take a peek.

John Jumper, Nobel Prize winner born after 1985

“A key to unlock the mysteries of the universe and life”

On October 9, Beijing time, the 2024 Nobel Prize in Chemistry was awarded to three scientists who made outstanding contributions to the design and prediction of protein structures. Half of the prize was awarded to American scientist David Baker in recognition of his contribution to "computational protein design"; the other half was awarded to two scientists working at the British artificial intelligence company Google DeepMind - Demis Hassabis and John M. Jumper (the former is the co-founder and CEO of DeepMind, and the latter is a senior research scientist at DeepMind) in recognition of their "contributions to protein structure prediction."

The Nobel Prize Committee commented that the three scientists "cracked the code of the amazing structure of proteins", which will bring great benefits to mankind. Baker successfully completed the "almost impossible task" of building a new protein, while Hassabis and Jumper developed an artificial intelligence model called AlphaFold2 that can predict the complex structure of proteins.

So, how do we know the importance of these studies? First of all, we have to talk about the role of protein.

Protein is a very important component of living things and the basis of all life activities.

Chemists have long dreamed of fully understanding and mastering proteins, the "chemical tools of life." But this dream is difficult to achieve. Because proteins are usually made of 20 amino acids twisted and folded like beads to form a unique three-dimensional shape, and the shape itself determines the function of the protein - some of them can create muscles, horns or feathers, while others may become hormones or antibodies. Many proteins form enzymes, which drive the chemical reactions of life with amazing precision. Proteins located on the surface of cells are also important. They act as communication channels between cells and their surroundings... Studying them is like cracking the code of life.

Since the 1970s, scientists have tried to predict the three-dimensional structure of a protein from its amino acid sequence, but this goal is as difficult as unlocking the mysteries of the universe. Although traditional experimental methods (such as X-ray crystallography, nuclear magnetic resonance spectroscopy, and cryo-electron microscopy) can provide high-resolution protein structure data, they are usually time-consuming, expensive, and technically challenging. In addition, these methods are not applicable to all proteins (such as membrane proteins), limiting their widespread application.

Just like a bright light suddenly lit up in the fog, four years ago, Hassabis and Jumper gave us a surprise - they launched an artificial intelligence model called AlphaFold2. This model is just as described in the evaluation of the Nobel Prize Committee: it solves a 50-year-old problem, can use AI to predict the complex structures of about 200 million known proteins, and has been used by more than 2 million people in 190 countries and regions around the world.

One of the "superheroes" who created this masterpiece, Jiang Pei, has attracted much attention because he is not only the main person in charge of the AlphaFold model, but also a young genius scientist. He was born in Little Rock, Arkansas, USA in 1985. Interestingly, Jiang Pei's initial curiosity about the universe drove him to study physics and mathematics. He taught himself computer programming in his teens and dreamed of "becoming a physicist who 'discovers the laws of the universe'". But when he began to use supercomputers to simulate proteins and their dynamics, he discovered that the physical knowledge of exploring the universe could be used to solve medical problems related to human health, just like using a magic key that not only opens the door to the universe, but also accidentally discovers another door to the mysteries of life.

Ushering in a new era of artificial intelligence cell biology

In 2017, after obtaining his doctorate, Jiang Po learned that DeepMind was conducting research on protein structure prediction. He submitted his application and soon became an important member of the team. At that time, AlphaFold developed by DeepMind had not made further breakthroughs, and Jiang Po proposed to use new ideas to improve the original design. Soon, he was promoted to AlphaFold project director and co-led the development of AlphaFold 2 with Hassabis.

Led by Jiang Po, a young team reorganized the original version of AlphaFold, made comprehensive adjustments and improvements on this basis, and explored every detail to achieve perfection. For example, they introduced spatial three-dimensional structure and evolutionary concepts; integrated detailed information on existing protein structures such as atomic radius and bond angles; improved machine learning strategies to extract the maximum information from limited data; and especially abandoned the constraints of traditional algorithms, emphasizing spatial proximity rather than linear proximity.

On July 15, 2021, a paper describing the content of AlphaFold2 was published online in the journal Nature, which immediately caused a sensation in the scientific community. This achievement was selected by Science as the top ten scientific breakthroughs of 2021. In that year, various "Top Ten Scientific Advances" around the world also listed AlphaFold as the top, which is rare in many years.

In 2021, DeepMind published a paper in Nature and open-sourced the AlphaFold 2 model

AlphaFold2 embodies Jiangpo's unique creativity. It can predict the three-dimensional structure of proteins directly from amino acid sequences through deep learning models, greatly improving the accuracy and efficiency of predictions. The average accuracy of its predictions reached 92.4%, which is almost comparable to the experimental results. It is worth mentioning that unlike the past when scientists usually took several years to obtain a protein structure, with the support of AI, the same experiment can now be completed in just a few minutes! This is really amazing.

At that time, scientists predicted that the 36-year-old Jiang Pei would win the Nobel Prize sooner or later, but many people did not expect the prediction to come true so soon. They also did not expect that he would win the Nobel Prize in Physiology or Medicine, or the Nobel Prize in Physics, but the Nobel Prize in Chemistry.

In May this year, DeepMind launched a new generation of products, AlphaFold 3, which can predict the structure of life molecules such as proteins, DNA, RNA, and small molecules with higher accuracy, including antibody-antigen interactions, which was difficult to achieve before. These advances make AlphaFold have a potentially "dramatic" impact in the field of biomedical research, which may accelerate the development of new drugs and innovation in the field of protein engineering, thus opening a new era of artificial intelligence cell biology.