Researchers at Meta, formerly Facebook, have developed a first-of-its-kind metagenomic database, the ESM Metagenomic Atlas, which could accelerate existing protein-folding AI performance by 60 times while maintaining accuracy. This impressive advancement makes Meta’s model more scalable to far larger databases.
The science and other stuff to know
Proteins are vital in keeping living organisms up and running. They help clear out waste, repair cells, and relay correspondences from one part of the body to another.
Over the years, scientists have done a great deal of work to decipher the structures and functions of proteins. In a new study published in the bioRxiv preprint server, it’s been reported that Meta’s AI research team has developed ESM Metagenomic Atlas, a model which can predict the 3D structure of proteins based on their amino acid sequences at a scale of hundreds of millions of proteins.
To achieve structure predictions at this scale, a breakthrough in the speed of protein folding is necessary. “What Meta is bringing to the table is a new protein-folding approach that harnesses large language models to create the first comprehensive view of the structures of proteins in a metagenomics database at the scale of hundreds of millions of proteins,” the research team wrote in a press release.
Unlike previous works, such as DeepMind’s AlphaFold, Meta’s AI is based on a language learning model rather than a shape-and-sequence matching algorithm. The scientists at Meta trained their model to recognize evolutionary patterns and produce precise structure predictions from start to finish from the protein sequence. More than 617 million protein structures were predicted in total. The effort only took two weeks to complete, and the predictions were 60 times faster than the state-of-the-art.
Meta’s model could be used in research applications such as understanding the function of a protein’s active site at the biochemical level. This could be very useful in drug development and discovery. Researchers also claim that Meta AI could even be used to design new proteins in the future.
“The ESM Metagenomic Atlas will enable scientists to search and analyze the structures of metagenomic proteins at the scale of hundreds of millions of proteins,” the team said. “This can help researchers to identify structures that have not been characterized before, search for distant evolutionary relationships, and discover new proteins that can be useful in medicine and other applications.”
Language-based prediction models are better suited to quickly determine how mutations alter protein structure. This isn’t possible with the shape-and-sequence matching algorithm in DeepMind’s technology.
“We will see structure prediction become leaner, simpler cheaper and that will open the door for new things,” Burkhard Rost, a computational biologist at the Technical University of Munich in Germany and who wasn’t involved in the study, said of Meta’s new AI.