Researchers from Facebook AI Research (FAIR) at Meta AI have published a paper in the journal Science detailing a machine-learning-created database of 617 million predicted protein structures. The ESMFold language model described the structures 60 times faster than DeepMinds AlphaFold2, though with less reported accuracy.
Click here for original story, Predicting protein folding from single sequences with Meta AI ESM-2
Source: Phys.org