The Nobel Prize is one of science’s most coveted honors. They are awarded annually to distinguished individuals for outstanding work that benefited humanity. This year, the Nobel Prize in Chemistry recognizes an achievement that forever changed the world of medicine.
The Nobel Prize in Chemistry: Hassabis, Baker, Jumper
Hassabis: “for computational protein design”
Baker and Jumper: “for protein structure prediction”
Photo Credit: Niklas Elmehed
Proteins are key to life as molecular tools to execute all functions in every living organism. Each protein has a unique structure encoded in the very sequence of amino acids. When proteins are first synthesized in the cell, they exist as only a strand of unfolded amino acids before folding up to adopt their 3-dimensional structure. This is called protein folding.
For years, scientists have been trying to deduce the structureof proteins. The first method used was X-ray crystallography. In 1912, German physicist Max von Laue and his colleagues discovered that after condensing a solution of proteins into a crystal and shining an X-ray through it, a diffraction pattern could be projected onto a surface behind it. X-rays can be diffracted due to the electrons in the atoms of the protein, and a map of electron density could be created. Using this map, scientists could match specific atoms with the electron density data and thus create a perfect model of a protein.
Throughout the 1900s, this process was used to determine protein structure, but it was painstakingly slow. Worldwide, only about one hundred protein structures could be mapped a year, much of the reason being the need to create an atomic model of a protein by hand. However, during the 1950s computers were used instead to create the protein models from X-ray crystallography data. This rapidly increased the efficiency of deducing protein structure, with a hundred proteins mapped per month.
The turn of the 20th century brought about the age of prediction of structure rather than determining structure known structures. Instead of deducing structures from proteins themselves, scientists looking to use computer models to predict protein structure from DNA sequences. In 1994, the Critical Assessment of Structure Prediction (CASP) was founded, a community-wide, worldwide experiment for protein structure prediction that takes place biannually. The challenge motivated scientists across the world to find solutions to the protein folding problem. Over the years, scientist gradually improved their CASP score, a percentage accuracy comparison between the computer-predicted model and the true structure. Over the years, this score hovered at a meager 50%, not nearly high enough to be considered for practical medical or scientific uses.
However, Nobel Prize winners in Chemistry in 2024 broke this trend — achieving a modeling accuracy of over 90% with their program called AlphaFold 2. Demis Hassabis, John Jumper from Google DeepMind, and David Baker from the University of California, Berkeley, joined forces to use deep learning and artificial intelligence to produce these results.
A diagram representing the algorithm for AlphaFold. Image credit: NJEM
Achieving this level of accuracyhas tremendous implications for today’s medical world. The efficiency of current medical and biological research requiring knowledge of protein structure would greatly increase. Researchers can determine the structure of proteins just by entering a DNA sequence. This saves valuable time without having to rely on X-ray crystallography to find the true structure.
Furthermore, AlphaFold can be used in drug discovery in a world where antibiotic resistance is rampant. Given the shape of the structure the protein should bind to, such as a receptor on a cell or another protein found in bacteria, Alphafold can create an artificial protein structure and move through the algorithm to give the new DNA sequence to produce such a protein. This DNA can then be inserted into bacteria to then mass produce proteins, thus creating new drugs.
