How to replicate results?

[123sjbsjb]

My error

Generate the ESM2 embeddings for the proteins:

python datasets/esm_embedding_preparation.py

I downloaded the BindingMOAD dataset and extracted it into the "/data/BindingMOAD_2020_ab_processed_biounit" directory.
I ran "datasets/esm_embedding_preparation.py", But file run error "NotADirectoryError: [Errno 20] Not a directory: '/mnt/data/sjb/chem/DiffDock/DiffDock-main//data/BindingMOAD_2020_ab_processed_biounit/pdb_protein/._1xq6_1_protein.pdb/ _1xq6_1_protein.pdb_protein.pdb'".

I modified the default code to use moad and found that after the MOAD dataset set was decompressed, some files started with '.', I modified line 73 of "esm_embedding_preparation.py". From "names = [n[:6] for n in names]" to "names = [n[:6] for n in names if not n.startswith('.')]", the error that the file cannot be found is no longer reported.

But there are new problems, the following is all the errors:

(diffdock) sjb@amax:/mnt/data/sjb/chem/DiffDock/DiffDock-main$ python datasets/esm_embedding_preparation.py
0% | | 4/54984 [00:00<23:40, 38.69 it/s] / home/SJB /. Conda/envs/diffdock/lib/python3.9 / site - packages/Bio/PDB/StructureBuilder py: 127: PDBConstructionWarning: WARNING: Residue (' ', 5, ' ') redefined at line 2353.
warnings.warn(
/ home/SJB /. Conda/envs/diffdock/lib/python3.9 / site - packages/Bio/PDB/StructureBuilder py: 149: PDBConstructionWarning: WARNING: Residue (' ', 5, ' ','SER') already defined with the same name at line  2353.
warnings.warn(
/ home/SJB /. Conda/envs/diffdock/lib/python3.9 / site - packages/Bio/PDB/PDBParser py: 340: PDBConstructionWarning: PDBConstructionException: Atom N defined twice in residue <Residue SER het=  resseq=5 icode= > at line 2353.
Exception ignored.
Some atoms or residues may be missing in the data structure.
warnings.warn(
/ home/SJB /. Conda/envs/diffdock/lib/python3.9 / site - packages/Bio/PDB/PDBParser py: 340: PDBConstructionWarning: PDBConstructionException: Atom CA defined twice in residue <Residue SER het=  resseq=5 icode= > at line 2355.
Exception ignored.
Some atoms or residues may be missing in the data structure.
warnings.warn(
/ home/SJB /. Conda/envs/diffdock/lib/python3.9 / site - packages/Bio/PDB/PDBParser py: 340: PDBConstructionWarning: PDBConstructionException: Atom C defined twice in residue <Residue SER het=  resseq=5 icode= > at line 2357.
Exception ignored.
Some atoms or residues may be missing in the data structure.
warnings.warn(
/ home/SJB /. Conda/envs/diffdock/lib/python3.9 / site - packages/Bio/PDB/PDBParser py: 340: PDBConstructionWarning: PDBConstructionException: Atom O defined twice in residue <Residue SER het=  resseq=5 icode= > at line 2359.
Exception ignored.
Traceback (most recent call last):
File "/mnt/data/sjb/chem/DiffDock/DiffDock-main/datasets/esm_embedding_preparation.py", line 88, in <module>
l = get_structure_from_file(rec_path)
File "/mnt/data/sjb/chem/DiffDock/DiffDock-main/datasets/esm_embedding_preparation.py", line 28,  in get_structure_from_file
structure = biopython_parser.get_structure('random_id', file_path)
The File "/ home/SJB /. Conda/envs/diffdock/lib/python3.9 / site - packages/Bio/PDB/PDBParser py", line 92, in get_structure
self._parse(lines)
The File "/ home/SJB /. Conda/envs/diffdock/lib/python3.9 / site - packages/Bio/PDB/PDBParser py", line 115, in _parse
self.trailer = self._parse_coordinates(coords_trailer)
The File "/ home/SJB /. Conda/envs/diffdock/lib/python3.9 / site - packages/Bio/PDB/PDBParser py", line 240, in _parse_coordinates
structure_builder.init_residue(
The File "/ home/SJB /. Conda/envs/diffdock/lib/python3.9 / site - packages/Bio/PDB/StructureBuilder py", line 177, in init_residue
self.chain.add(self.residue)
The File "/ home/SJB /. Conda envs/diffdock/lib/python3.9 / site - packages/Bio/PDB/Entity. Py", line 215, in the add
raise PDBConstructionException("%s defined twice" % entity_id)
TypeError: not all arguments converted during string formatting

Dear Author How should this be resolved?If you can answer, thank you very much for your help

[123sjbsjb]

This seems to be an issue with the biopython package. I checked the "biopython/Bio/PDB/Entity. py" section of biopython.
They has been changed from

PDBConstructionException ("% s defined twice"% entity_id)

to

If self. has_id (entity_id):
Raise PDBConstructionException (f "{entity_id} defined twice")

I upgraded the version of biopython from1.76 to 1.83, although it can run, the MOAD dataset still displays a large number of "{entity_id} defined twices". This is completely different from the PDBbind dataset. Why is this?