Model of human nuclear pore complex, built using AlphaFold2; credit: Agnieszka Obarska-Kosinska, Nature
AlphaFold 2: A breakthrough in computational protein folding
The 2024 Nobel Prize in Chemistry has been awarded to David Baker of the University of Washington and to Demis Hassabis and John Jumper of Google DeepMind. Hassabis and Jumper, now at the Google DeepMind Research Center in the U.K. (a subsidiary of Alphabet, Google’s parent company), developed an artificial intelligence-based software package that predicts with remarkable accuracy the structure of proteins. Baker, now at the University of Washington in Seattle, was recognized for his work on designing new proteins.
Demis Hassabis and John Jumper are the leading researchers behind AlphaFold 2, the computer-based system developed at DeepMind that has largely solved the computational protein folding problem, which is arguably one of the most significant and longest-running challenges in modern biology — certainly the most significant in the area of computational biology.
Proteins are the workhorses of biology. A few prominent examples include:
- Actin, myosin, tubulin and keratin: proteins that form the structure of the human body and enable muscles to work.
- Hemoglobin: the basis of red blood that carries oxygen to cells throughout the body.
- Amylase, lipase, pepsin and trypsin: key proteins involved in digestion.
- Insulin and thyroxine: key hormones in the regulation of metabolism.
- Spike glycoprotein: the spike-shaped protein that enables the coronavirus to invade healthy cells.
There are many thousands of proteins in human biology, and at least 200 million in the larger biological kingdom. Each protein is specified as a string of amino acids, as given by DNA letters A, C, T, G. Thanks to the recent dramatic drop in the cost of DNA sequencing technology, sequencing proteins is fairly routine.
The key to biology, however, is the three-dimensional shape of the protein — how a protein “folds” [Callaway2022]. Protein shapes can be investigated experimentally, using x-ray crystallography, but this is an expensive, error-prone and time-consuming laboratory operation. So for at least 50 years, researchers worldwide have been developing computer programs to model the folding process, given the input amino acid sequence [Dill2008].
Just a few of the many potential applications of this technology include studying the misshapen proteins thought to be the cause of Alzheimer’s disease, or studying the proteins behind various genetically-linked disorders such as cystic fibrosis and sickle-cell anemia. Prior to the development of AlphaFold 2, structures were known for only a tiny fraction of known proteins.
Given the daunting challenge and importance of the computational protein folding problem, in 1994 a community of researchers in the field organized a biennial competition known as Critical Assessment of Protein Structure Prediction (CASP) [CASP2024]. At each iteration of the competition, the organizers announce a set of problems, to which worldwide teams of researchers then apply their best current tools to solve. In 2018, the CASP competition had a new entry: AlphaFold, a machine-learning-based program developed by DeepMind, which out-performed others in the competition. Then in 2020, the DeepMind team unveiled a new program, known as AlphaFold 2 [AlphaFold 2]. For the 2020 CASP competition, it achieved a 92% average score, far above the 62% achieved by the second-best program in the competition.
The Nobel committee notes that with AlphaFold 2, the DeepMind team first calculated the structure of all known human proteins, and then extended their work to determine “the structure of virtually all the 200 million proteins that researchers have so far discovered when mapping Earth’s organisms.”
“It’s a game changer,” exulted German biologist Andrei Lupas, who has served as an organizer and judge for the CASP competition. “This will change medicine. It will change research. It will change bioengineering. It will change everything.” [Callaway2020]. Lupas mentioned how AlphaFold 2 helped to crack the structure of a bacterial protein that Lupas himself has been studying for many years. “The [AlphaFold 2] model … gave us our structure in half an hour, after we had spent a decade trying everything.”
For additional information, see the Nobel committee announcement and articles in Nature, New Scientist, New York Times and Scientific American.
Hassabis and Jumper were previously (2022) honored with the Breakthrough Prize for their work; for details, see this earlier Math Scholar article.