John Jumper
Last reviewed
Jun 5, 2026
Sources
24 citations
Review status
Source-backed
Revision
v2 ยท 2,125 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Suggest edit
CtrlK
Last reviewed
Jun 5, 2026
Sources
24 citations
Review status
Source-backed
Revision
v2 ยท 2,125 words
Add missing citations, update stale details, or suggest a clearer explanation.
John Jumper (born 1 January 1985) is an American computational chemist and biophysicist and a senior researcher at Google DeepMind, where he led the development of AlphaFold, the artificial intelligence system that predicts the three-dimensional structure of proteins from their amino acid sequences. In 2024 he shared the Nobel Prize in Chemistry with Demis Hassabis and David Baker, with Jumper and Hassabis cited "for protein structure prediction." [1][2][12]
Jumper is best known for solving a problem that had occupied structural biologists for roughly half a century: how to determine a protein's folded shape directly from its sequence. Trained as a theoretical and computational chemist, he brought together physics, statistics, and machine learning to build AlphaFold2, which at the 2020 CASP14 assessment predicted protein structures with an accuracy approaching that of experimental methods. [2][3] The system and the open database built from it have been used by more than two million researchers worldwide, and the 2021 paper describing AlphaFold became one of the most-cited scientific publications of its era. [1][4]
Jumper was born on 1 January 1985 in Little Rock, Arkansas. [1][5][12] He attended Pulaski Academy in Little Rock, graduating in 2003, and has described himself as an "accidental chemist" who had not taken a chemistry class since high school before later building his career in the field. [5][13] He earned a Bachelor of Science in physics and mathematics from Vanderbilt University in 2007. [5][6] On a Marshall Scholarship he then studied at the University of Cambridge, where as a member of St Edmund's College he completed a Master of Philosophy in theoretical condensed matter physics in 2010, working on numerical methods for quantum Monte Carlo calculations of small molecules. [5][7][14]
He moved to the University of Chicago for graduate study, receiving a master's degree in theoretical chemistry in 2012 and a PhD in theoretical chemistry in 2017. [5][8] His doctoral research, supervised by Tobin Sosnick and Karl Freed, applied machine learning to coarse-grained models of protein folding and dynamics, learning simulation parameters directly from the Protein Data Bank. [7][8] Jumper has recalled that he entered the chemistry program with essentially no formal background in the subject, saying "When I started, I knew no chemistry. None. I had to speedrun the whole thing." [13] The University of Chicago later noted that Jumper was the 100th scholar affiliated with the university to be associated with a Nobel Prize. [8]
Before completing his doctorate, Jumper worked for about three years at D. E. Shaw Research in New York City, the computational biochemistry laboratory founded by David E. Shaw. There he developed and ran molecular dynamics simulations of proteins and other molecules, including studies of supercooled liquids, using purpose-built high-performance computing. [5][7][15] That experience in large-scale physical simulation of biomolecules shaped his later conviction that machine learning could be combined with structural data to predict folding outcomes. [7] He has said that writing computational protein-modeling programs at the company captivated him and drew him toward biology, noting how directly the work connected to human outcomes. [13]
Jumper joined DeepMind in 2017 as a research scientist, arriving in time to contribute ideas to the first version of AlphaFold, and was promoted in July 2018 to lead the development of AlphaFold2. [5][7][16] He was the lead author of the 2021 Nature paper "Highly accurate protein structure prediction with AlphaFold," published on 15 July 2021 (Nature, volume 596, pages 583 to 589), which set out the architecture and results of AlphaFold2. [1][9] At DeepMind he later led the development of AlphaFold3, an updated diffusion-based model announced in May 2024 that predicts the joint structure of proteins together with DNA, RNA, small-molecule ligands, and ions. [7][10][17]
DeepMind describes Jumper as a director, a role he has held since 2023, while the Royal Society lists his title as Distinguished Scientist at Google DeepMind. [2][7][15] He has consistently characterized AlphaFold as the achievement of a large team, saying after the Nobel announcement that the award "recognizes their amazing work." [2]
Jumper's published work centers on a sequence of AlphaFold systems and the structural database built from their predictions. His Google Scholar profile records more than 100,000 total citations, reflecting the speed with which the work was taken up across biology. [18]
| Year | Paper | Venue | What it introduced | Approx. citations |
|---|---|---|---|---|
| 2020 | Improved protein structure prediction using potentials from deep learning (AlphaFold 1) | Nature | Deep residual network predicting distances between residue pairs [16] | ~4,400 [18] |
| 2021 | Highly accurate protein structure prediction with AlphaFold (AlphaFold2) | Nature | Evoformer attention architecture plus a structure module for end-to-end 3D prediction [1][9] | ~50,000 [18] |
| 2021 | Protein complex prediction with AlphaFold-Multimer | bioRxiv | Extension of AlphaFold2 to multi-chain protein complexes [18] | ~4,100 [18] |
| 2022 | AlphaFold Protein Structure Database: massively expanding structural coverage | Nucleic Acids Research | Open database of predicted structures across known proteomes [4][18] | ~8,800 [18] |
| 2024 | Accurate structure prediction of biomolecular interactions with AlphaFold 3 | Nature | Pairformer module and a diffusion-based generator for proteins, nucleic acids, ligands, and ions [17] | ~15,000 [18] |
The first AlphaFold system debuted at the CASP13 assessment in late 2018, where it placed first. It used a deep residual neural network to predict probability distributions over the distances between pairs of amino acid residues, and Andrew Senior, Richard Evans, John Jumper, James Kirkpatrick, and Laurent Sifre are credited as joint first authors of that work. [16] AlphaFold2, which Jumper led, replaced that pipeline with a new neural network design. Its Evoformer block jointly reasons over a multiple sequence alignment and a representation of residue pairs, and a downstream structure module converts those representations directly into atomic coordinates. [3][9] AlphaFold3, on which Jumper is the senior author and Josh Abramson is the lead author, was published in Nature in 2024 (volume 630, pages 493 to 500). It replaces much of the earlier machinery with a Pairformer module and a diffusion model that starts from a cloud of atoms and iteratively refines their positions, allowing it to predict the joint structure of proteins together with other classes of molecule. [17]
The structure-prediction problem rests on a paradox. In the 1960s Christian Anfinsen showed that a protein's three-dimensional structure is determined by its amino acid sequence, while Cyrus Levinthal pointed out that a protein could in principle adopt an astronomical number of conformations, so that folding by random search would take longer than the age of the universe. Yet cells fold proteins in milliseconds. [3] The Critical Assessment of Structure Prediction (CASP) competition, launched in 1994, measured progress on predicting structures from sequence, but for years accuracy stalled. [3]
AlphaFold2 changed that. At CASP14 in 2020, judges found that for most targets the system predicted structures with an accuracy comparable to experimental techniques such as X-ray crystallography, a result the Nobel committee called "astounding." [3] In the assessors' own analysis the system reached a median domain accuracy of 92.4 on the Global Distance Test, and in the ranking by summed z-scores it scored 244.0 against 90.8 for the next best group, a margin that set it well apart from the rest of the field. [19] A distinctive feature of the model is that it reports a calibrated confidence estimate for each prediction, telling users which regions to trust. [7]
After releasing the AlphaFold code and methods publicly, DeepMind and its collaborator EMBL's European Bioinformatics Institute built the AlphaFold Protein Structure Database, which expanded to cover roughly 200 million proteins, close to all cataloged proteins known to science. [3][4] By late 2024 the database and tools had been used by more than two million people in 190 countries, supporting work in drug discovery, enzyme engineering, vaccine design, and basic biology, and shortening tasks that once took years to a matter of minutes. [3][4]
On 9 October 2024 the Royal Swedish Academy of Sciences awarded the Nobel Prize in Chemistry to David Baker "for computational protein design" and jointly to Demis Hassabis and John Jumper "for protein structure prediction." [11] Baker received one half of the prize; Hassabis and Jumper shared the other half, so Jumper's share was one quarter. [11][12] The Academy framed the prizes as recognizing two related advances: the design of entirely new proteins and the fulfillment of "a 50-year-old dream" of predicting protein structures from their amino acid sequences. [11]
The Nobel citation specifically honored AlphaFold2, the 2020 model with which Hassabis and Jumper "have been able to predict the structure of virtually all the 200 million proteins that researchers have identified." [11] Jumper, then 39, was among the youngest chemistry laureates in decades. [5][11] He delivered his Nobel lecture in Stockholm on 8 December 2024 and received the prize at the award ceremony at the Stockholm Concert Hall on 10 December 2024. [20]
In addition to the Nobel Prize, Jumper has received several of the major awards in the life sciences, often jointly with Demis Hassabis or other members of the AlphaFold team.
| Year | Award | Awarding body |
|---|---|---|
| 2021 | Named to Nature's 10 | Nature [5] |
| 2022 | Wiley Prize in Biomedical Sciences | Wiley Foundation [5][21] |
| 2022 | VinFuture Special Prize | VinFuture Foundation [5][22] |
| 2022 | Frontiers of Knowledge Award in Biology and Biomedicine | BBVA Foundation [5][23] |
| 2023 | Breakthrough Prize in Life Sciences | Breakthrough Prize Foundation [5][7] |
| 2023 | Canada Gairdner International Award | Gairdner Foundation [5][7] |
| 2023 | Albert Lasker Award for Basic Medical Research | Lasker Foundation [5][7] |
| 2024 | Nobel Prize in Chemistry | Royal Swedish Academy of Sciences [11] |
| 2025 | Golden Plate Award | American Academy of Achievement [5] |
| 2025 | Marshall Medal | Marshall Aid Commemoration Commission [24] |
| 2025 | Fellow of the Royal Society | Royal Society [7] |
| 2026 | Member, National Academy of Engineering | National Academy of Engineering [5] |
He was elected a Fellow of the Royal Society in 2025, with the citation crediting him for leading the development of the AlphaFold2 and AlphaFold3 systems and for developing "entirely new machine learning architectures" that demonstrated structure-prediction systems could provide accurate confidence measurements. [7] The Marshall Aid Commemoration Commission awarded him the Marshall Medal in 2025, recognizing the impact of work that began during his time as a Marshall Scholar at Cambridge. [24] In 2026 he was elected a member of the National Academy of Engineering. [5]
As of 2026 Jumper continues to work at Google DeepMind in London, where he leads research on protein structure prediction and related applications of machine learning to biology. [2][7] His group's work has remained central to DeepMind's scientific program and to its drug-discovery spinout Isomorphic Labs, which co-developed AlphaFold3. [10][17]