It says the University of Washington. I kinda wonder if we're just basically in a hamster wheel and they call us the A.I.
Why is helix the only thing that scores high? I always throw some sheets in my designs just because. Or why can't long loops get in there?
Here is a nice overview at Nature of both the new UW model (RoseTTAFold) and the public release of the AlphaFold2 model: DeepMind’s AI for protein structure is coming to the masses
UW's paper on RoseTTAFold: Accurate prediction of protein structures and interactions using a three-track neural network
DeepMind's paper on AlphaFold2: Highly accurate protein structure prediction with AlphaFold
Here is a lengthy article from WIRED on the race between UW and DeepMind to make the latest structure prediction methods available for scientists to build on: Without Code for DeepMind’s Protein AI, This Lab Wrote Its Own
Indeed one of the reviewers (in their blog) pointed out that:
"it was humorous to see people suggest that the open sourcing of AF2 was in response to RoseTTAFold—it was in fact DeepMind’s plan well before RoseTTAFold was preprinted."