Senior Research Scientist at Deepmind
In 2007, John Jumper received degree of Bachelor of Science in Physics and Mathematics at Vanderbilt University, USA. He has studied adaptive time-step methods for quantum Monte Carlo at University of Cambridge in 2008. After that, he worked on molecular dynamics simulations of proteins and supercooled liquids at D. E. Shaw Research for 3 years. He started his Ph.D. program in 2011 and received the degree of Doctor of Philosophy in Theoretical Chemistry at University of Chicago in 2017. Since 2018, he has worked as a senior research scientist at Deepmind, London, UK. The scientific journal Nature included John Jumper as one of the ten “people who mattered” in science in their annual listing of Nature’s 10 2021.
In 2022, he received the Wiley Prize in Biomedical Sciences. He received the 2023 Breakthrough Prize in Life Sciences for developing AlphaFold, which accurately predicts the structure of protein.
Proteins do most of the work in cells and are required for the structure, function, and regulation of all body’s tissues and organs. All living things on Earth are made from proteins including animals, bacteria, plants, yeast, virus and so on. They are simple strings of amino acids that fold up from a linear chain into complex, compact 3D shapes. A protein can fold in a near infinite number of ways before reaching its final structure which determines the functions of the protein. A tiny change in the structure can dramatically change the functions of a protein. Therefore, determining the folding structures of the proteins are super important to understand their functions and more importantly, generate the suggestions for drug discovery. The structure of protein was thought to be determined by its amino acid sequence. But proving that was a whole different ball game, and the protein folding problem has been a headache that has plagued and puzzled scientists for 50 years. With the team DeepMind, Jumper and Hassabis conceived and constructed a novel machine learning approach that incorporates physical and biological knowledge about protein structure, leveraging multi-sequence alignments, into the design of the deep learning algorithm, that accurately and rapidly models the structure of proteins.
With AlphaFold, AI (artificial intelligence) was proved that it could accurately predict the shape of a protein down to atomic accuracy, at scale and in minutes, AlphaFold not only provided a solution to a 50-year grand challenge, but it also became the first big proof point of our founding thesis: that artificial intelligence can dramatically accelerate scientific discovery, and in turn advance humanity. AlphaFold is accelerating drug discovery by providing a better understanding of newly identified proteins that could be drug targets, and helping scientists to more quickly find potential medicines that bind to them. The AlphaFold DB serves as a ‘google search’ for protein structures, providing researchers with instant access to predicted models of the proteins they’re studying, enabling them to focus their effort and expedite experimental work.
A system that can accurately predict protein structure has an enormous potential on benefiting many areas. It can help increase our understanding on neglected tropical diseases, the unknown protein structure of which is a barrier of curing them. Such diseases affect millions of people every year, causing tens of thousands of deaths. Due to limited information about human involved protein structures, the process of drug discovery is slow and costly, taking about 10 years and 2,5 billion dollars. In addition, a system like AlphaFold could help finding proteins and enzymes that can break down industrial plastic waste or efficiently capturing carbon from the atmosphere. Thus, the potential of using the protein structure for braking down industrial plastic waste is very promising on assisting the global need for reducing the global material footprint.
Just after 12 months, AlphaFold has been accessed by more than half a million researchers and used to accelerate progress on important real-world problems ranging from plastic pollution to antibiotic resistance. The database of AlphaFold today has 300+ million of protein structures (nearly every known protein from across the tree of life) while the experimental protein database has only 190,000 structures. AlphaFold is free for everyone. Jumper and Hassabis decided to open sourced AlphaFold’s code and published two in-depth articles in Nature, which have already been cited more than 4000 times in just more than a year. To date, more than 500,000 researchers from 190 countries have accessed the AlphaFold database to view over 2 million structures.
Traditionally, research has relied on expensive and time-consuming methods to work out structures, such as X-ray crystallography and electron microscopy. It can take from a few months to several years for a biologist to crack the puzzle. “Even then, success is not guaranteed – some proteins are notoriously difficult to find structures of” says Pushmeet Kohli, head of AI for science at DeepMind. With AlphaFold, any researcher will be able to get a protein structure in mere minutes. The Universal Protein database, a collection of all the proteins that science has uncovered thus far, contains over 180 million protein sequences. These protein sequences tell us how the amino acids in a protein are ordered. But, to really understand how proteins function in the body, we need to know how that sequence determines the 3D structure of the protein – and that is a much more difficult task than simply knowing the right order of amino acids. Of those 180 million protein sequences, scientists have so far worked out the structure of just 180,000 proteins. AlphaFold provides predictions for more than double the number of known protein structures to date. Now biologists will be able to work on understanding how proteins interact and function – and beyond that, designing new proteins, enabling quicker drug discovery, deciphering disease-causing gene variations and more.