Open keiran-rowell-unsw opened 4 hours ago
https://github.com/Australian-Structural-Biology-Computing/proteinfold/blob/9af55b19269bca2fb9de8565ef1b71cac7d2a279/modules/local/run_alphafold2.nf#L68
Is it worth checking to see if the Amber/OpenMM relaxation has failed, and it so re-try from _unrelaxed.pdb with a CPU relaxation.
It's slower, but the reason this work is the double precision of the CPU can deal with gradient changes to get you out of a tricky spot.
https://github.com/Australian-Structural-Biology-Computing/proteinfold/blob/9af55b19269bca2fb9de8565ef1b71cac7d2a279/modules/local/run_alphafold2.nf#L68
Is it worth checking to see if the Amber/OpenMM relaxation has failed, and it so re-try from _unrelaxed.pdb with a CPU relaxation.
It's slower, but the reason this work is the double precision of the CPU can deal with gradient changes to get you out of a tricky spot.