Dr. John M. Jumper > CV

Predicting protein structure has long been a challenging area of biomolecular science. All twenty amino acids were identified in 1935 and the first protein structures, haemoglobin and myoglobin, were solved in 1958 by Max Perutz and John Kendrew (for which they were awarded a Nobel Prize in Chemistry in 1962). Yet predicting three-dimensional protein structures from amino acid sequences remained a cumbersome task, with a large margin for error.

In 2016, a team at DeepMind took on the challenge of identifying protein structure. The initial version of AlphaFold, developed by DeepMind, achieved 60% accuracy, which was far better than previous attempts, but still short of the required 90% experimental accuracy.

This was changed by a recent PhD graduate, who had joined DeepMind as senior research scientist. John M. Jumper brought his experience of working on protein simulation and led the work on AlphaFold2 (in addition to doing research himself). The AI model uses machine learning, as well as knowledge from biology and physics to identify proteins with similar amino acid sequence and characteristics – which amino acid chains could be attracted to one another and which could be repelled. The neural network-based model succeeded in predicting protein structure with 90% accuracy (on average). Within 18 months of its introduction, AlphaFold2 predicted the structure of all known 200 million proteins and has become an indispensable tool in structural and molecular biology, as well as protein engineering.

“We’re closing an enormous gap, where protein sequences are discovered about 3000 times faster than protein structure. We need machine learning to close that gap,” said Jumper during his Nobel lecture.

John Michael Jumper was born in Little Rock, Arkansas in 1985. He earned a B.A. in mathematics and physics from Vanderbilt University in 2007. He then graduated from the University of Cambridge (2008) with an MPhil degree in theoretical condensed matter physics, carrying out his research at the Cavendish Laboratory. Jumper left academia for three years to work at D.E. Shaw, where he focused on the computer simulation of proteins. He was awarded a PhD from the University of Chicago in 2017 for his thesis, “New Methods Using Rigorous Machine Learning for Coarse-Grained Protein Folding and Dynamics.” Jumper remained at the university as a postdoc for nearly a year before moving to Google DeepMind, where he is now (2025) a Distinguished Scientist.