Graylab / IgFold

Fast, accurate antibody structure prediction from deep learning on massive set of natural antibodies
Other
319 stars 60 forks source link

fix the align_mask bug when calculating the prmsd_loss #49

Open yaowensnf opened 1 year ago

yaowensnf commented 1 year ago

The shape of atom-wise align_mask is (batch, seq_len*4), need to multiply 4 to the indexes when masking out residues.

qiz218591 commented 11 months ago

Hi, I am getting error, this ValueError Traceback (most recent call last) c:\Users\s4735779\Documents\GitHub\IgFold\IgFold.ipynb Cell 6 line 1 10 igfold = IgFoldRunner(num_models=num_models) 12 pred_pdb = os.path.join(pred_dir, f"{name}.pdb") ---> 13 pred = igfold.fold( 14 pred_pdb, 15 sequences=sequences, 16 do_refine=do_refine, 17 use_openmm=True, 18 do_renum=do_renum, 19 ) 20 show_pdb(pred_pdb, len(sequences), bb_sticks=False, sc_sticks=True, color="rainbow")

File c:\Users\s4735779\Documents\GitHub\IgFold\igfold\IgFoldRunner.py:106, in IgFoldRunner.fold(self, pdb_file, fasta_file, sequences, template_pdb, ignore_cdrs, ignore_chain, skip_pdb, do_refine, use_openmm, do_renum, truncate_sequences) 91 """ 92 Predict antibody structure with IgFold. 93 (...) 103 :param truncate_sequences: Truncate sequences with AbNumber. 104 """ 105 start_time = time() --> 106 model_out = fold( 107 self.antiberty, 108 self.models, ... (...) 228 do_renum=do_renum, 229 )

ValueError: min() arg is an empty sequence Output is truncated. View as a scrollable element or open in a text editor. Adjust cell output settings...

why?? please solve