search

jwohlwend / boltz

Official repository for the Boltz-1 biomolecular interaction model
MIT License
1.28k stars 129 forks source link

Truncated final protein chains #70

Closed fregat17 closed 2 days ago

fregat17 commented 2 days ago

Hi team, thank you for your impressive work.

Unfortunately, I encountered a problem when generating homooligomeric complexes. The monomer length is 641 amino acids, but when I add more of the same chains, the final structure has only 219 amino acids per chain, and there are UNK residues after each chain (One UNK after each 219AA chain). This behavior is observed in di-, tri-, and tetramer complexes when viewed in PyMOL.

I'm using version 0.3.0, with an input .yaml file containing a list of chains [A, B, C, D] in the same protein entity and a 1xV100 GPU, if that matters.

jwohlwend commented 2 days ago

Hi, could you share your input YAML?

fregat17 commented 2 days ago

Sorry for disturbing, actually my mistake