Meta’s AI unlocks the power of metagenomics to accelerate protein discovery

Meta’s latest AI can determine proper protein folding 60 times faster

The life we know today would not exist if it weren’t for the proteins molecules. They enable everything from photosynthesis to enzymatic degrading, and even our sight and immune system. As with most aspects of nature, humans have only recently begun to explore the vast array of proteins that exist. Meta researchers developed the ESM Metagenomic Atlas – a metagenomic database that is the first of its kind. It could improve the performance of existing AIs in protein folding by 60x.

Metagenomics was named by accident. Metagenomics is just a coincidence.

The NCBI, European Bioinformatics Institute and Joint Genome Institute have all launched similar databases that have cataloged millions of new protein shapes. Meta’s contribution is \”a novel protein-folding method that uses large language models in order to create a comprehensive view of the structure of proteins within a metagenomics dataset at a scale of hundreds and millions of proteins,\” as stated in a company release on Tuesday. While genomic advances have revealed sequences for many novel proteins, they don’t tell us exactly how these sequences fit together to form a functional molecule. It can take anywhere between a few months and a few year to figure it out experimentally. Each molecule. No one has time for that.