For nearly 50 years, scientists have struggled to solve one of nature's most perplexing challenges — predicting the complex 3D shape a string of amino acids will twist and fold into as it becomes a fully functional protein. This year, scientists have shown that artificial intelligence (AI)-driven software can achieve this long-standing goal and predict accurate protein structures by the thousands and at a fraction of the time and cost involved with previous methods.
To honor this feat, Science has named AI-powered protein prediction as its 2021 Breakthrough of the Year.
"This is a breakthrough that will keep on giving. Being able to predict protein shapes isn't just a decades-old dream come true," said Tim Appenzeller, the lead editor for Science's news section. "It will also transform future work, from understanding the workings of the cell to developing new drugs."
Proteins are the building blocks of life, and their myriad functions are central to nearly all biological processes. Each is born a linear chain of amino acids, assembled through the instructions encoded in our DNA, and subsequently folded into elaborate 3D shapes, which determine how they interact with other molecules and define their biological functions. It's estimated that nearly 200 million proteins exist across all life forms — as many as 400,000 in the human body alone.
Given their importance to basic biological function and their immense potential in aiding drug development, the ability to determine a protein's shape represents an important achievement.
To date, determining a protein's structure was a time-consuming and costly process, involving complicated lab analyses and methods like X-ray crystallography and cryo-electron microscopy (cryo-EM). Experimental characterization of the shape of a single protein with these methods can take years and cost upwards of hundreds of thousands of dollars.
While attempts to develop computer-aided models capable of solving the "protein-folding problem" have been ongoing for decades, accurate protein prediction has eluded researchers until two seminal papers, published nearly simultaneously in Nature and Science in July, presented AlphaFold2 and RoseTTA-fold, respectively. Each AI-driven software program demonstrates the ability to predict a wide variety of complex protein structures quickly and accurately based solely on the amino acids they contain.
AlphaFold, developed by the Google sister company DeepMind, trained itself on databases of solved protein structures and showed protein structure prediction at accuracies comparable to current experimental techniques. RoseTTA fold — developed by a team led by David Baker, a University of Washington computational biochemist, performed similarly to AlphaFold, but using only a fraction of the computational processing power and time. With RoseTTA-fold, a protein structure can be computed in as little as ten minutes on a single, commercially available desktop computer.
What's more, both groups made their data and code freely available to researchers, greatly expanding the accessibility of obtaining protein structures.
In the time since, both AI approaches have revealed thousands of new protein structures, including those for nearly half of those found in the human body.
"This is a breakthrough on two fronts," writes Science Editor-in-Chief Holden Thorp in a related Editorial. "First, it solves a scientific problem that has been on the to-do list for 50 years… Second, it's a game-changing technique that, like CRISPR or cryo-EM, will greatly accelerate scientific discovery."
Breakthrough Runners Up
Runners up for the Breakthrough of the Year include the development of antiviral pills to fight COVID-19; new measurements of the muon; seismic observations from Mars; recovering ancient human DNA from soils; application of CRISPR gene editing inside the body; new insights into early human development; use of psychedelic drugs to treat PTSD; development of lab-made monoclonal antibodies for treating infectious disease; and important advancements in fusion energy generation.
Breakdowns of the Year
Science also named several "Breakdowns" of the year. They include the botched U.S. Food and Drug Administration approval of the new Alzheimer's drug, aducanumab; the growing realization that the hope of limiting global warming to 1.5°C is becoming unlikely; and the public animosity against scientists, doctors and public health officials due to the politicization of the COVID-19 response.
[credit for associated images: Minkyung Baek & AAAS; Science/AAAS]