Replies
CASP14
#s
just came out and they’re astounding—DeepMind looks to have solved protein structure prediction. Median GDT_TS went from 68.5 (CASP13) to 92.4!!!! Cf. their 2nd best CASP13 struct scored 92.8 (out of 100). Median RMSD is 2.1Å. I think it's over
35
565
2K
These are for single domains-not whole proteins-and there are a few poor predictions. So corner cases remain but core problem appears solved: 88% of predictions are <4Å, 76% <3Å, 46% <2Å. Unlike last time where there was some competition, this time AF2 was best for 88/97 targets.
6
19
171
DeepMind's blogpost:
3
43
194
Nature's write-up:
2
77
311
@MarioNiepel
They haven’t described their method yet but we’ll find out more today. It does appear that they have residue-level error metrics so the answer to your question is most likely yes.
1
0
0