DeepMind uncovers the structure of 200m proteins in a scientific leap forward: Every known protein has been decrypted by artificial intelligence, paving the way for new treatments or technology that can help solve global problems like famine or pollution.
Proteins are the basic building blocks of all living things on the planet. The 3D structure mainly determines their function, as they are made up of chains of amino acids folded into complicated forms. In order to understand how proteins function, you first need to know how they fold up. Until recently, scientists had only deciphered a portion of the 200 million or so proteins known to science, despite the fact that DNA supplies the instructions for creating the chain of amino acids.
DeepMind uncovers the structure of 200m proteins in a scientific leap forward
DeepMind, an AI company, stated in November 2020 that it had built a tool called AlphaFold that could anticipate this information quickly using an algorithm. Every organism that has had its genome sequenced since then has been crunching through the genetic instructions and anticipating the architectures of hundreds of millions of proteins.
Earlier this year, DeepMind published the structures of proteins from 20 species, including nearly all of the 20,000 proteins that are expressed in humans. It has now completed its task and released more than 200 million predicted protein structures.
“You can conceive of it as encompassing the entire protein cosmos. AlphaFold will be able to have a significant influence on crucial topics like sustainability, food insecurity, and neglected diseases because it incorporates prediction structures for plants, bacteria, mammals, and many more creatures,” stated Demis Hassabis, DeepMind’s founder and CEO.
Some of its previous forecasts are already being used by scientists in the development of new treatments. Researchers at the University of Oxford, led by Professor Matthew Higgins, reported in May that they had utilized AlphaFold models to help determine the structure of a critical malaria parasite protein and determine where antibodies that could inhibit parasite transmission were likely to bind.
In the past, Higgins and his colleagues tried to figure out what this molecule looked like using a process called protein crystallography, but they couldn’t get a handle on it since it’s so dynamic. “Suddenly, everything made sense when we integrated the AlphaFold models with this experimental evidence. In the future, this information will be utilized to build more effective vaccines that limit the transmission of disease.
This approach is also being utilized by scientists at Portsmouth’s Centre for Enzyme Innovation, who are looking for natural ingredients that may be modified to break down and recycle plastics. According to project leader Prof. John McGeehan, “it took us quite a long time to dig through this vast database of structures but opened this whole palette of novel three-dimensional geometries we’d never seen before that could actually break down plastics. This is an entirely new way of thinking.” We have the ability to accelerate our progress from here on out, which allows us to better allocate our limited resources to the things that really matter.”
AlphaFold protein structure predictions are already being used in a variety of ways, according to Prof Dame Janet Thornton, a senior scientist at the European Molecular Biology Laboratory. In the months and years to come, I fully expect this latest upgrade to unleash an avalanche of new and fascinating discoveries, thanks to the data being freely available to all.”
