DeepMind built an algorithm called AlphaFold, which by analyzing the chemical makeup of thousands of known proteins and their 3D shapes, has used that information to predict the shapes of unknown proteins with high accuracy. In July they announced that they have used AlphaFold to predict the structure of 214 million proteins from more than one million species, essentially all known protein-coding sequences, and make them publicly available for free.
The prediction of a protein’s structure from its DNA sequence alone has been one of biology’s greatest challenges. Current experimental methods to determine the shape of a single protein take months or years in a laboratory This is why it is estimated that only about 190,000, or 0.1%, of known protein structures, have been solved.
In July 2021, DeepMind announced it had predicted the shape of all human proteins, helping to better understand human health and diseases. These structures along with the +200m structural models just announced are available in an open database jointly maintained with the European Molecular Biology Laboratory’s European Bioinformatics Institute.
Over the past year, scientists have applied AlphaFold in all sorts of ways. Some have used its predictions to identify new families of proteins (which now need to be verified experimentally). Some are using it to help the search for drugs to treat neglected diseases. AlphaFold’s current predictions on their own are not a monumental step change in drug discovery they are working on an evolution of the AI which can predict how proteins change shape when they interact with each other.