To highlight major inflection points in research, the Vanderbilt School of Medicine Basic Sciences launched the Apex Lecture Series earlier this year, which allows the Basic Sciences community to engage with researchers from around the world who are influencing the trajectory of their fields.
John Jumper, BS’07, senior staff research scientist for the artificial intelligence company DeepMind, will give the next Apex Lecture on August 30 at 4:00 p.m. CDT in Light Hall room 208. Jumper’s lecture is co-sponsored by the recently launched Vanderbilt Center for Applied Artificial Intelligence in Protein Dynamics.
Jumper leads the team that developed AlphaFold, a trailblazing artificial intelligence system that can accurately predict the 3D structures of proteins. AlphaFold2, with over 200 million protein structures predicted, has created a generational launching point for expanding structural biology and its applications to impact human health.
The study of protein structures is integral to targeting diseases. “Almost everything that happens in life is based on proteins serving as the worker bees,” stated Chuck Sanders, vice dean of the School of Medicine Basic Sciences and president of the Protein Society. “When things go right, they go right because your proteins are working correctly, and when things go wrong, it is usually because something is wrong with one or more of your proteins. This is why 99% of all drugs target proteins.”
To understand how proteins in the body function, interact, and bind to molecules for drug discovery, protein structures must be visualized. Historically, this has been done through processes such as x-ray crystallography and, more recently, cryogenic electron microscopy, both of which help determine the 3D structures of molecules. However, these methods are expensive and can take weeks or months to obtain a clear structure. AlphaFold has revolutionized how protein structure research is done by greatly increasing the accuracy of protein structure prediction and providing vast amounts data that allow for the analysis of protein interactions with potential therapeutics and improvements on the mechanistic research on proteins.
“Throughout my scientific journey, I have been fascinated with the protein folding problem,” stated Hassane Mchaourab, director of the Center for Applied Artificial Intelligence in Protein Dynamics. “I always expected that the answer to decoding how a protein sequence dictates its structure will be accomplished through computer algorithms that apply the laws of physics. Jumper and his colleagues showed how a deep learning algorithm can learn the energetics of protein folding from evolution and examples of structures. It is part of the AI tsunami that is reshaping society in unparalleled ways, from knowledge dissemination to the practice of medicine.”
Jumper’s unique education allowed him to become a pioneer in the protein structure field. He graduated from Vanderbilt with a degree in physics and mathematics and then studied at the University of Cambridge, receiving an M.Phil. in theoretical condensed matter physics. Jumper then worked as a scientific associate at D.E. Shaw Research, where he performed research on molecular dynamics using computer simulations and developed an algorithm to extract key data from these simulations. In 2017, he earned a Ph.D. in theoretical chemistry from the University of Chicago with his thesis titled “New methods using rigorous machine learning for coarse-grained protein folding and dynamics.” Before joining DeepMind, Jumper completed a postdoctoral fellowship at University of Chicago, continuing his work with deep learning models for protein prediction.
For his innovative work on AlphaFold and its impacts on biomedical research, Jumper has received numerous awards, including the 2023 Breakthrough Prize in Life Sciences.
“We are excited to welcome John Jumper back to Vanderbilt for this Apex Lecture,” stated John Kuriyan, dean of basic sciences. “AlphaFold represents a revolutionary advance in structural biology, one that has brought the holy grail of predictive protein folding into the toolkit of biochemistry and molecular biology. The impact of these advances on drug discovery and, ultimately, on human health, will be enormous.”
Lecture information:
Jumper will present his lecture, “Highly Accurate Protein Structure Prediction and Its Applications,” on August 30 at 4:00pm CDT in Light Hall room 208. A reception will follow in the Light Hall lobby.
Abstract: Our work on deep learning, specifically the AlphaFold system, has demonstrated that neural networks are capable of highly accurate modeling of both protein structure and protein-protein interactions. In particular, the system shows a remarkable ability to extract chemical and evolutionary principles from experimental structural data. This computational tool has repeatedly shown the ability to not only predict accurate structures for novel sequences and novel folds but also to do unexpected tasks such as selecting stable protein designs or detecting protein disorder. In this lecture, I will discuss the context of this breakthrough in the machine learning principles, the diverse data and rigorous evaluation environment that enabled it to occur, and the many innovative ways in which the community is using these tools to do new types of science. I will also reflect on some surprising limitations — insensitivity to mutations and the lack of context about the chemical environment of the proteins — and how this may be traced back to the essential features of the training process. Finally, I will conclude with a discussion of some ideas on the future of machine learning in structure biology and how the experimental and computational communities can think about organizing their research and data to enable many more such breakthroughs in the future.