How to inference with custome PDBs?

[ELMIAR-0642]

How to run the provided Inference code with custom set of Antibody and Antigen PDBs?

Should it be preprocessed to be in DB5 format? How to do that?

[yaledeus]

You can adopt following steps to construct your custom test set:

1. your directory should contain two folders named as complexes and random_transformed and two files named as test.json and test.txt
2. test.txt: each line shows the name of one complex, e.g., 1AVX
3. test.json: each line is a json format dictionary, containing following keys: PDB, which should be the same as in test.txt, e.g., 1AVX; rchain, the chain names of the receptor, list format, e.g., ["A"]; lchain, the chain names of the ligand.
4. complexes: for each ground-truth complex, split it into two PDB files named as {name}_l_b_COMPLEX.pdb and {name}_r_b_COMPLEX.pdb, representing the ligand and receptor respectively. Here {name} should be the same as in test.txt.
5. random_transformed: for each complex, split it into two PDB files named as {name}_l_b.pdb and {name}_r_b.pdb, while at least one of them should be transformed by a random rotation and translation.
6. Put your directory under test_sets_pdb, add your dataset name into the argument --dataset, and insert codes into test.py to support loading from your dataset, like this:

   if args.dataset == 'DB5':
       test_path = './test_sets_pdb/db5_test_random_transformed'
   ...
   elif args.dataset == {your_dataset_name}:
       test_path = './test_sets_pdb/{your_dataset_directory_name}'
       test_desc_path = os.path.join(test_path, 'test.json')
       with open(test_desc_path, 'r') as fin:
           lines = fin.read().strip().split('\n')
       for line in lines:
           item = json.loads(line)
           test_desc[item['pdb']] = [item['rchain'], item['lchain']]

[ELMIAR-0642]

Can't Access VM rn, will check it and let you know.

Thanks to you. You have been so responsive.

[ELMIAR-0642]

(ellidock2) root@2d175a2198f1:/ElliDock# python test.py --dataset custom_set_000 --gpu 0 --ckpt /ElliDock/checkpoints/db5_be
st.ckpt

Model type: ElliDock 0%| | 0/1 [00:00<?, ?it/s] Traceback (most recent call last): File "test.py", line 272, in main(parse()) File "test.py", line 119, in main batch = test_complex_process(ligand_bound_path, receptor_unbound_path) File "/ElliDock/data/dataset.py", line 683, in test_complex_process complex = BaseComplex.from_pdb(ligand_path, receptor_path) File "/ElliDock/data/bio_parse.py", line 313, in from_pdb ligand_seq, ligand_coord = base_pdb_parse(ligand_path) File "/ElliDock/data/bio_parse.py", line 60, in base_pdb_parse structure = parser.get_structure(pdb_id, pdb_path) File "/conda/envs/ellidock2/lib/python3.8/site-packages/Bio/PDB/PDBParser.py", line 89, in get_structure with as_handle(file) as handle: File "/conda/envs/ellidock2/lib/python3.8/contextlib.py", line 113, in enter return next(self.gen) File "/conda/envs/ellidock2/lib/python3.8/site-packages/Bio/File.py", line 72, in as_handle with open(handleish, mode, **kwargs) as fp: FileNotFoundError: [Errno 2] No such file or directory: './test_sets_pdb/custom_set_000/complexes/1AVX_l_b_COMPLEX.pdb'

I don't wish to compare complexes and docked ( like Ground Truth vs Predicted Truth) but to dock the receptor and ligand

[koushikmj]

Facing a similar issue. I was interested in using this for a project. But I am not able to find out how to run inference for docking 2 proteins.

[koushikmj]

(ellidock2) root@2d175a2198f1:/ElliDock# python test.py --dataset custom_set_000 --gpu 0 --ckpt /ElliDock/checkpoints/db5_be
st.ckpt

Model type: ElliDock 0%| | 0/1 [00:00<?, ?it/s] Traceback (most recent call last): File "test.py", line 272, in main(parse()) File "test.py", line 119, in main batch = test_complex_process(ligand_bound_path, receptor_unbound_path) File "/ElliDock/data/dataset.py", line 683, in test_complex_process complex = BaseComplex.from_pdb(ligand_path, receptor_path) File "/ElliDock/data/bio_parse.py", line 313, in from_pdb ligand_seq, ligand_coord = base_pdb_parse(ligand_path) File "/ElliDock/data/bio_parse.py", line 60, in base_pdb_parse structure = parser.get_structure(pdb_id, pdb_path) File "/conda/envs/ellidock2/lib/python3.8/site-packages/Bio/PDB/PDBParser.py", line 89, in get_structure with as_handle(file) as handle: File "/conda/envs/ellidock2/lib/python3.8/contextlib.py", line 113, in enter return next(self.gen) File "/conda/envs/ellidock2/lib/python3.8/site-packages/Bio/File.py", line 72, in as_handle with open(handleish, mode, **kwargs) as fp: FileNotFoundError: [Errno 2] No such file or directory: './test_sets_pdb/custom_set_000/complexes/1AVX_l_b_COMPLEX.pdb'

I don't wish to compare complexes and docked ( like Ground Truth vs Predicted Truth) but to dock the receptor and ligand

Yo, did you find any solution of your own?

[joelBrelson]

Even i am facing similar kind of issue has any one found a soultion for this. I badly need this for my college project.

[yaledeus]

If you only want to generate docked complex structures without evaluation, we have provided inference.py similar to test.py, and you can follow the command in README:

Now you do not have to construct complexes directory for your own test set any more.