Merge pull request #65 from UCLOrengoGroup/add-pdb-globularity

Add globularity measure for generic PDB files
UCLOrengoGroup · Nov 15, 2023 · 56ced3c · 56ced3c
2 parents b262dea + 8ae2bec
commit 56ced3c
Show file tree

Hide file tree

Showing 9 changed files with 598 additions and 235 deletions.
diff --git a/cath_alphaflow/cli.py b/cath_alphaflow/cli.py
@@ -17,6 +17,7 @@
 from .commands import convert_cif_to_fasta
 from .commands import load_mongo
 from .commands import measure_globularity
+from .commands import measure_globularity_af_cif
 
 logging.basicConfig(
     level=logging.INFO, format="%(asctime)s | %(levelname)s | %(message)s"
@@ -67,3 +68,4 @@ def dump_config():
 cli.add_command(convert_cif_to_fasta.convert_cif_to_fasta)
 cli.add_command(load_mongo.load_af_from_archive)
 cli.add_command(measure_globularity.measure_globularity)
+cli.add_command(measure_globularity_af_cif.measure_globularity_af_cif)
diff --git a/cath_alphaflow/commands/create_dataset_cath_files.py b/cath_alphaflow/commands/create_dataset_cath_files.py
@@ -15,6 +15,7 @@
     Gene3DCrhPredictedCathDomainProvider,
     DecoratedCrhPredictedCathDomainProvider,
 )
+from pydantic import ConfigDict
 
 LOG = logging.getLogger()
 
@@ -128,12 +129,12 @@ def create_cath_dataset_from_files(src_decorated_crh, **kwargs):
 
 class CathDatasetGeneratorBase(pydantic.BaseModel):
 
-    csv_uniprot_ids: typing.Any  # click.File
-    gene3d_crh_output: typing.Any  # click.File
-    csv_uniprot_md5: typing.Any  # click.File
-    af_domainlist_ids: typing.Any  # click.File
-    af_chainlist_ids: typing.Any  # click.File
-    af_cath_annotations: typing.Any  # click.File
+    csv_uniprot_ids: typing.Any = None  # click.File
+    gene3d_crh_output: typing.Any = None  # click.File
+    csv_uniprot_md5: typing.Any = None  # click.File
+    af_domainlist_ids: typing.Any = None  # click.File
+    af_chainlist_ids: typing.Any = None  # click.File
+    af_cath_annotations: typing.Any = None  # click.File
     chunk_size: int
     af_version: int
     fragment_number: int
@@ -142,9 +143,7 @@ class CathDatasetGeneratorBase(pydantic.BaseModel):
     af_domainlist_writer: typing.Any = None
     af_chainlist_writer: typing.Any = None
     af_cath_annotations_writer: typing.Any = None
-
-    class Config:
-        arbitrary_types_allowed = True
+    model_config = ConfigDict(arbitrary_types_allowed=True)
 
     def run(self):
 

diff --git a/cath_alphaflow/commands/load_mongo.py b/cath_alphaflow/commands/load_mongo.py
@@ -18,6 +18,7 @@
 from cath_alphaflow.models.mongo import AFFile, AFFileType
 from cath_alphaflow.models import beacons
 from cath_alphaflow.errors import ParseError
+from pydantic import ConfigDict
 
 LOG = logging.getLogger(__name__)
 
@@ -119,9 +120,7 @@ class AFArchiveFile(pydantic.BaseModel):
 
     filename: str
     tarfileobj: tarfile.ExFileObject
-
-    class Config:  # pylint: disable=missing-class-docstring
-        arbitrary_types_allowed = True
+    model_config = ConfigDict(arbitrary_types_allowed=True)
 
 
 def yield_next_file_from_archive(archive_path) -> AFArchiveFile: