chore: hydra and pip

apps/protein_folding/helixfold3/helixfold/common/all_atom_pdb_save.py

@@ -164,10 +164,13 @@ def prediction_to_mmcif(pred_atom_pos: Union[np.ndarray, paddle.Tensor],
     - maxit_binary: path to maxit_binary, use to convert pdb to cif
     - mmcif_path: path to save *.cif
   """
-  assert maxit_binary is not None and os.path.exists(maxit_binary), (
+  if os.path.isfile(maxit_binary):
+    raise FileNotFoundError(
       f'maxit_binary: {maxit_binary} not exists. '
       f'link: https://sw-tools.rcsb.org/apps/MAXIT/source.html')
-  assert mmcif_path.endswith('.cif'), f'mmcif_path should endswith .cif; got {mmcif_path}'
+
+  if not mmcif_path.endswith('.cif'):
+     raise ValueError(f'mmcif_path should endswith .cif; got {mmcif_path}')
 
   pdb_path = mmcif_path.replace('.cif', '.pdb')
   pdb_path = prediction_to_pdb(pred_atom_pos, FeatsDict, pdb_path)

apps/protein_folding/helixfold3/helixfold/config/helixfold.yaml

@@ -0,0 +1,59 @@
+defaults:
+  - _self_
+
+# General configuration
+
+bf16_infer: false  # Corresponds to --bf16_infer
+seed: null  # Corresponds to --seed
+logging_level: DEBUG  # Corresponds to --logging_level
+job_id: 'structure_prediction'  # Corresponds to --model_name
+weight_path: /mnt/db/weights/helixfold/HelixFold3-params-240814/HelixFold3-240814.pdparams  # Corresponds to --init_model
+precision: fp32  # Corresponds to --precision
+amp_level: O1  # Corresponds to --amp_level
+infer_times: 1  # Corresponds to --infer_times
+diff_batch_size: -1  # Corresponds to --diff_batch_size
+use_small_bfd: false # Corresponds to --use_small_bfd
+
+# File paths
+
+input: null  # Corresponds to --input_json, required field
+output: null  # Corresponds to --output_dir, required field
+
+
+# Binary tool paths, leave them as null to find proper ones under PATH or conda bin path
+bin:
+  jackhmmer: null    # Corresponds to --jackhmmer_binary_path
+  hhblits: null  # Corresponds to --hhblits_binary_path
+  hhsearch: null   # Corresponds to --hhsearch_binary_path
+  kalign: null  # Corresponds to --kalign_binary_path
+  hmmsearch: null  # Corresponds to --hmmsearch_binary_path
+  hmmbuild: null  # Corresponds to --hmmbuild_binary_path
+  nhmmer: null  # Corresponds to --nhmmer_binary_path
+  obabel: null
+
+# Database paths
+db:
+  uniprot: /mnt/db/uniprot/uniprot.fasta  # Corresponds to --uniprot_database_path, required field
+  pdb_seqres: /mnt/db/pdb_seqres/pdb_seqres.txt  # Corresponds to --pdb_seqres_database_path, required field
+  uniref90: /mnt/db/uniref90/uniref90.fasta  # Corresponds to --uniref90_database_path, required field
+  mgnify: /mnt/db/mgnify/mgy_clusters.fa  # Corresponds to --mgnify_database_path, required field
+  bfd: /mnt/db/bfd/bfd_metaclust_clu_complete_id30_c90_final_seq.sorted_opt  # Corresponds to --bfd_database_path
+  small_bfd: null  # Corresponds to --small_bfd_database_path
+  uniclust30: /mnt/db/uniref30_uc30/UniRef30_2022_02/UniRef30_2022_02  # Corresponds to --uniclust30_database_path
+  rfam: /mnt/db/helixfold/rna/Rfam-14.9_rep_seq.fasta  # Corresponds to --rfam_database_path, required field
+  ccd_preprocessed: /mnt/db/ccd/ccd_preprocessed_etkdg.pkl.gz  # Corresponds to --ccd_preprocessed_path, required field
+
+# Template and PDB information
+template:
+  mmcif_dir: /mnt/db/pdb_mmcif/mmcif_files  # Corresponds to --template_mmcif_dir, required field
+  max_date: '2023-03-15'  # Corresponds to --max_template_date, required field
+  obsolete_pdbs: /mnt/db/pdb_mmcif/obsolete.dat  # Corresponds to --obsolete_pdbs_path, required field
+
+# Preset configuration
+preset:
+  preset: full_dbs  # Corresponds to --preset, choices=['reduced_dbs', 'full_dbs']
+
+# Other configurations
+other:
+  maxit_binary: /mnt/data/yinying/software/maxit/maxit-v11.100-prod-src/bin/maxit  # Corresponds to --maxit_binary
+  no_msa_templ_feats: false  # Corresponds to --no_msa_templ_feats

apps/protein_folding/helixfold3/infer_scripts/preprocess.py → apps/protein_folding/helixfold3/helixfold/infer_scripts/preprocess.py

@@ -5,17 +5,17 @@
         'seqs': ccd_seqs,
         'msa_seqs': msa_seqs,
         'count': count,
-        'extra_mol_infos': {}， for which seqs has the modify residue type or smiles.
+        'extra_mol_infos': {}, for which seqs has the modify residue type or smiles.
 """
 import collections
 import copy
+import gzip
 import os
 import json
 import sys
 import subprocess
 import tempfile
 import itertools
-sys.path.append('../')
 import rdkit
 from rdkit import Chem
 from rdkit.Chem import AllChem
@@ -52,9 +52,7 @@
     3: 'Unknown error.'
 }
 
-OBABEL_BIN = os.getenv('OBABEL_BIN')
-if not os.path.exists(OBABEL_BIN):
-    raise FileNotFoundError(f'Cannot find obabel binary at {OBABEL_BIN}.')
+
 
 
 def read_json(path):
@@ -144,6 +142,11 @@ def smiles_toMol_obabel(smiles):
     """
         generate mol from smiles using obabel;
     """    
+
+    OBABEL_BIN = os.getenv('OBABEL_BIN')
+    if not (OBABEL_BIN and os.path.isfile(OBABEL_BIN)):
+        raise FileNotFoundError(f'Cannot find obabel binary at {OBABEL_BIN}.')
+
     with tempfile.NamedTemporaryFile(suffix=".mol2") as temp_file:
         print(f"[OBABEL] Temporary file created: {temp_file.name}")
         obabel_cmd = f"{OBABEL_BIN} -:'{smiles}' -omol2 -O{temp_file.name} --gen3d"