Machine learning solves protein folding problem and wins 2024 Nobel Prize in Chemistry
The 2024 Nobel Prize in Chemistry will be awarded to Demis Hassabis and John Jumper for using machine learning to tackle one of biology’s biggest challenges: predicting the 3D shape of proteins and designing them from scratch. , honored David Baker.
This year’s awards were notable because they celebrated research originating from technology companies, namely DeepMind, an AI research startup acquired by Google in 2014. To date, most Nobel Prizes in Chemistry have been awarded to researchers in academia. Many laureates have gone on to found start-up companies to further scale and commercialize their breakthrough research, such as CRISPR gene editing technology and quantum dots, but their research is conducted in the commercial realm from start to finish. It’s not like I was hurt.
Although the Nobel Prize in Physics and Chemistry are awarded separately, there is an interesting link between the research awarded in these fields in 2024. The Physics Prize went to two computer scientists who laid the foundations for machine learning, while the Chemistry Prize winners received: They used machine learning to tackle one of biology’s biggest mysteries: how proteins fold.
The 2024 Nobel Prizes highlight both the importance of this type of artificial intelligence and how science today often crosses traditional boundaries and fuses different disciplines to achieve breakthrough results. I’m doing it.
Challenge to protein folding
Proteins are the molecular machines of life. They make up important parts of our bodies, including muscles, enzymes, hormones, blood, hair, and cartilage.
Understanding protein structure is essential because a protein’s shape determines its function. In 1972, Christian Anfinsen won the Nobel Prize in Chemistry for showing that the sequence of a protein’s amino acid building blocks determines the protein’s shape, which in turn influences its function. If proteins are not folded correctly, they may not function properly and can lead to diseases such as Alzheimer’s disease, cystic fibrosis, and diabetes.
The overall shape of a protein is determined by the small interactions, attractions and repulsions, between all the atoms of the amino acids that make it up. Some people want to be with you, some people don’t. Proteins twist and fold into their final shape based on thousands of these chemical interactions.
For decades, one of biology’s biggest challenges has been predicting the shape of proteins based solely on their amino acid sequences. Although researchers can now predict shapes, it is still unclear how proteins move into a particular shape and minimize the repulsion of all atomic interactions within microseconds. Not yet.
To understand how proteins work and prevent misfolding, scientists needed a way to predict how proteins would fold, but solving this puzzle is a simple task. It wasn’t.
In 2003, University of Washington biochemist David Baker created Rosetta, a computer program for designing proteins. With this, he showed that it is possible to reverse the protein folding problem by designing the protein’s shape and predicting the amino acid sequence needed to create it.
Although this was an amazing leap forward, the geometry chosen for the calculations was simple and the calculations complex. Routinely designing novel proteins with desired structures required a major paradigm shift.
A new era of machine learning
Machine learning is a type of AI in which computers learn how to solve problems by analyzing vast amounts of data. It’s used in a variety of fields, from gameplay and speech recognition to self-driving cars and scientific research. The idea behind machine learning is to use patterns hidden in data to answer complex questions.
This approach took a big leap forward in 2010 when Demis Hassabis co-founded DeepMind, a company that aims to combine neuroscience and AI to solve real-world problems.
Hassabis, a four-year-old chess prodigy, quickly became a hit with AlphaZero, an AI that learned to play chess at a superhuman level. In 2017, AlphaZero completely defeated Stockfish-8, the world’s top computer chess program. AI’s ability to learn from its own gameplay rather than relying on pre-programmed strategies has marked a turning point in the world of AI.
Soon after, DeepMind applied a similar technique to Go, an ancient board game known for its great complexity. In 2016, the company’s AI program AlphaGo defeated Lee Sedol, one of the world’s top players, in a widely watched match that surprised millions of people.
In 2016, Hassabis shifted DeepMind’s focus to a new challenge: protein folding. The AlphaFold project began under the leadership of John Jumper, a chemist with a background in protein science. The researchers trained the AI ​​using a large database of experimentally determined protein structures, allowing it to learn the principles of protein folding. As a result, AlphaFold2, an AI that can predict the three-dimensional structure of proteins from amino acid sequences with surprising accuracy, was born.
This was an important scientific advance. Since then, AlphaFold has predicted the structures of over 200 million proteins. This is basically every protein that scientists have ever sequenced. This huge database of protein structures is now freely available, accelerating research in biology, medicine, and drug discovery.
Designer proteins to fight disease
Understanding how proteins fold and function is critical to designing new drugs. Enzymes, a type of protein, act as catalysts for biochemical reactions and can speed up or regulate these processes. To treat diseases such as cancer or diabetes, researchers often target specific enzymes involved in disease pathways. By predicting the shape of a protein, scientists can understand where small molecules, potential drug candidates, are likely to bind to the protein. This is the first step in designing a new drug.
In 2024, DeepMind launched AlphaFold3, an upgraded version of the AlphaFold program that not only predicts protein shapes but also identifies potential binding sites for small molecules. This advancement makes it easier for researchers to design drugs that precisely target the right proteins.
Google reportedly acquired Deepmind in 2014 for about $500 million. Google DeepMind is now launching a new venture, Isomorphic Labs, to collaborate with pharmaceutical companies on real-world drug development using these AlphaFold3 predictions.
David Baker continues to make significant contributions to protein science. His team at the University of Washington developed an AI-based technique called “whole family illusion” and used it to design entirely new proteins from scratch. A hallucination is a plausible new pattern (in this case, a protein), meaning it closely matches the pattern in the AI’s training data. These new proteins include luminescent enzymes, demonstrating that machine learning can help create new synthetic proteins. These AI tools provide new ways to design functional enzymes and other proteins that could not have evolved naturally.
AI enables the next chapter of research
Hassabis, Jumper, and Baker’s Nobel Prize-worthy work shows that machine learning is no longer just a tool for computer scientists, but is now an essential part of the future of biology and medicine.
By tackling one of the most difficult problems in biology, the 2024 laureates are opening up new possibilities in drug discovery, personalized medicine, and even understanding the chemistry of life itself.
Provided by The Conversation
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Citation: Machine learning solves protein folding problem, wins 2024 Nobel Prize in Chemistry (October 12, 2024) https://phys.org/news/2024-10-machine-protein-problem-won Retrieved October 13, 2024 from -nobel.html
This document is subject to copyright. No part may be reproduced without written permission, except in fair dealing for personal study or research purposes. Content is provided for informational purposes only.