search

Graylab / IgFold

Fast, accurate antibody structure prediction from deep learning on massive set of natural antibodies
Other
325 stars 61 forks source link

Chain breaks after Rosetta refinement #52

Closed ta-rom closed 1 year ago

ta-rom commented 1 year ago

Hello!

Thank you for your great instrument.

I faced a problem with IgFold: sometimes Rosetta refinement creates breaks in the antibody chain.

A probable reason is that in the pdb_string of the predicted structure TER records are placed not at the end of the chain, but inside it (while they are in place after the write_pdb_bfactor function). I believe it is because delims list is determined as containing the number of the last residue in the chain, but later atom number (k) and not residue number (r) is searched in it.

jeffreyruffolo commented 1 year ago

Thanks for catching this! I believe I've now addressed the issue in a new release, v0.3.3. Please reopen if the issue persists though.