chore: Migrate cache models to pydantic v2

BALSAMIC/commands/init/base.py

@@ -128,7 +128,10 @@ def initialize(
         references=references,
         references_date=datetime.now().strftime("%Y-%m-%d %H:%M"),
     )
-    write_json(json.loads(cache_config.json(exclude_none=True)), config_path.as_posix())
+    write_json(
+        json_obj=json.loads(cache_config.model_dump_json(exclude_none=True)),
+        path=config_path.as_posix(),
+    )
     LOG.info(f"Reference workflow configured successfully ({config_path.as_posix()})")
 
     snakefile: Path = (

BALSAMIC/models/cache.py

@@ -1,9 +1,16 @@
 """Balsamic reference cache models."""
 import logging
 from pathlib import Path
-from typing import Dict, Optional, List, Any, Union
+from typing import Dict, Optional, List, Union
 
-from pydantic.v1 import BaseModel, AnyUrl, DirectoryPath, validator, FilePath
+from pydantic import (
+    BaseModel,
+    AnyUrl,
+    DirectoryPath,
+    FilePath,
+    field_validator,
+    ValidationInfo,
+)
 
 from BALSAMIC.constants.cache import GenomeVersion, GRCHVersion
 from BALSAMIC.constants.constants import FileType, BwaIndexFileType
@@ -31,8 +38,8 @@ class ReferenceUrl(BaseModel):
     gzip: bool
     file_name: str
     dir_name: str
-    file_path: Optional[str]
-    secret: Optional[str]
+    file_path: Optional[str] = None
+    secret: Optional[str] = None
 
 
 class References(BaseModel):
@@ -277,15 +284,15 @@ class CacheConfig(BaseModel):
     variants_dir: Path
     vep_dir: Path
     genome_version: GenomeVersion
-    cosmic_key: Optional[str]
+    cosmic_key: Optional[str] = None
     bioinfo_tools: dict
     containers: Dict[str, str]
     references: Union[ReferencesHg, ReferencesCanFam]
     references_date: str
 
-    @validator("references")
+    @field_validator("references")
     def validate_references(
-        cls, references: References, values: Dict[str, Any]
+        cls, references: References, info: ValidationInfo
     ) -> References:
         """Validate the reference output paths."""
         for model in references:
@@ -294,15 +301,15 @@ def validate_references(
             reference_key, reference = model[0], model[1]
             reference.file_path = (
                 Path(
-                    values.get("references_dir"),
+                    info.data.get("references_dir"),
                     reference.dir_name,
                     reference.file_name,
                 ).as_posix()
                 if reference
                 else None
             )
             reference.secret = (
-                values.get("cosmic_key") if "cosmic" in reference_key else None
+                info.data.get("cosmic_key") if "cosmic" in reference_key else None
             )
         return references
 

BALSAMIC/workflows/reference.smk

@@ -17,7 +17,7 @@ from BALSAMIC.utils.utils import get_relative_paths_dict
 LOG = logging.getLogger(__name__)
 
 # Balsamic cache configuration model
-cache_config: CacheConfig = CacheConfig.parse_obj(config)
+cache_config: CacheConfig = CacheConfig.model_validate(config)
 
 # Temporary directory and shell options
 os.environ["TMPDIR"] = cache_config.references_dir.as_posix()
@@ -46,7 +46,7 @@ rule all:
     run:
         analysis_references: Dict[str, str] = get_relative_paths_dict(
             base_path=cache_config.references_dir,
-            data=cache_config.get_analysis_references().dict(),
+            data=cache_config.get_analysis_references().model_dump(),
         )
         write_json(
             json_obj=analysis_references,

CHANGELOG.rst

@@ -49,6 +49,7 @@ Changed:
 * Update `reference.json` file to use relative paths https://github.com/Clinical-Genomics/BALSAMIC/pull/1251
 * Update pydantic to v2 while maintaining support for v1 models https://github.com/Clinical-Genomics/BALSAMIC/pull/1253
 * `PCT_PF_READS_IMPROPER_PAIRS` QC threshold lowered to 5% https://github.com/Clinical-Genomics/BALSAMIC/issues/1265
+* Migrate cache models to pydantic v2 https://github.com/Clinical-Genomics/BALSAMIC/pull/1277
 
 Fixed:
 ^^^^^^

tests/conftest.py

@@ -1,5 +1,4 @@
 import copy
-import json
 import os
 import shutil
 from datetime import datetime
@@ -11,6 +10,7 @@
 import pytest
 from _pytest.tmpdir import TempPathFactory
 from click.testing import CliRunner
+from pydantic_core import Url
 
 from BALSAMIC import __version__ as balsamic_version
 from BALSAMIC.commands.base import cli
@@ -2043,9 +2043,9 @@ def fixture_analysis_references_hg(
 
 
 @pytest.fixture(scope="session", name="reference_url")
-def fixture_reference_url() -> str:
+def fixture_reference_url() -> Url:
     """Return dummy reference url."""
-    return "gs://gatk-legacy-bundles/b37/reference.vcf.gz"
+    return Url("gs://gatk-legacy-bundles/b37/reference.vcf.gz")
 
 
 @pytest.fixture(scope="session", name="reference_file")
@@ -2058,7 +2058,7 @@ def fixture_reference_file(session_tmp_path: Path) -> Path:
 
 @pytest.fixture(scope="session", name="reference_url_data")
 def fixture_reference_url_data(
-    reference_url: str, reference_file: Path, cosmic_key: str
+    reference_url: Url, reference_file: Path, cosmic_key: str
 ) -> Dict[str, Any]:
     """return reference url model data."""
     return {

tests/models/test_cache_models.py

@@ -4,7 +4,7 @@
 
 import pytest
 from _pytest.logging import LogCaptureFixture
-from pydantic.v1 import ValidationError
+from pydantic import ValidationError
 
 from BALSAMIC.constants.cache import (
     GRCHVersion,
@@ -35,7 +35,7 @@ def test_analysis_references(analysis_references_data: Dict[str, Path]):
     model: AnalysisReferences = AnalysisReferences(**analysis_references_data)
 
     # THEN the model should have been correctly built
-    assert model.dict() == analysis_references_data
+    assert model.model_dump() == analysis_references_data
 
 
 def test_analysis_references_empty():
@@ -60,7 +60,7 @@ def test_analysis_references_canfam(analysis_references_data: Dict[str, Path]):
     )
 
     # THEN the model should have been correctly built
-    assert model.dict() == analysis_references_data
+    assert model.model_dump() == analysis_references_data
 
 
 def test_analysis_references_canfam_empty():
@@ -83,7 +83,7 @@ def test_analysis_references_hg(analysis_references_hg_data: Dict[str, Path]):
     model: AnalysisReferencesHg = AnalysisReferencesHg(**analysis_references_hg_data)
 
     # THEN the model should have been correctly built
-    assert model.dict() == analysis_references_hg_data
+    assert model.model_dump() == analysis_references_hg_data
 
 
 def test_analysis_references_hg_empty():
@@ -106,7 +106,7 @@ def test_reference_url(reference_url_data: Dict[str, Any]):
     model: ReferenceUrl = ReferenceUrl(**reference_url_data)
 
     # THEN the model should have been correctly built
-    assert model.dict() == reference_url_data
+    assert model.model_dump() == reference_url_data
 
 
 def test_reference_url_empty():
@@ -120,7 +120,7 @@ def test_reference_url_empty():
         ReferenceUrl()
 
 
-def test_references(references_data: Dict[str, dict]):
+def test_references(references_data: Dict[str, dict], references: References):
     """Test references model."""
 
     # GIVEN an input for the reference model
@@ -129,7 +129,7 @@ def test_references(references_data: Dict[str, dict]):
     model: References = References(**references_data)
 
     # THEN the model should have been correctly built
-    assert model.dict() == references_data
+    assert model == references
 
 
 def test_references_empty():
@@ -219,7 +219,7 @@ def test_get_refgene_bed_file_path(references: References, refgene_bed_file: Pat
     assert refgene_output_file == refgene_bed_file.as_posix()
 
 
-def test_references_canfam(references_data: Dict[str, dict]):
+def test_references_canfam(references_data: Dict[str, dict], references: References):
     """Test canine references model."""
 
     # GIVEN an input for the canine reference model
@@ -228,7 +228,7 @@ def test_references_canfam(references_data: Dict[str, dict]):
     model: ReferencesCanFam = ReferencesCanFam(**references_data)
 
     # THEN the model should have been correctly built
-    assert model.dict() == references_data
+    assert model.model_dump() == references.model_dump()
 
 
 def test_references_canfam_empty():
@@ -242,7 +242,9 @@ def test_references_canfam_empty():
         ReferencesCanFam()
 
 
-def test_references_hg(references_hg_data: Dict[str, dict]):
+def test_references_hg(
+    references_hg_data: Dict[str, dict], references_hg: ReferencesHg
+):
     """Test human genome references model."""
 
     # GIVEN an input for the human genome reference model
@@ -251,7 +253,7 @@ def test_references_hg(references_hg_data: Dict[str, dict]):
     model: ReferencesHg = ReferencesHg(**references_hg_data)
 
     # THEN the model should have been correctly built
-    assert model.dict() == references_hg_data
+    assert model == references_hg
 
 
 def test_references_hg_empty():
@@ -353,7 +355,7 @@ def test_cache_analysis(cache_analysis_data: Dict[str, str]):
     model: CacheAnalysis = CacheAnalysis(**cache_analysis_data)
 
     # THEN the model should have been correctly built
-    assert model.dict() == cache_analysis_data
+    assert model.model_dump() == cache_analysis_data
 
 
 def test_cache_analysis_empty():
@@ -591,6 +593,7 @@ def test_get_reference_output_paths(cache_config: CacheConfig):
 def test_get_analysis_references_hg(
     cache_config: CacheConfig,
     analysis_references_hg_data: Dict[str, Path],
+    analysis_references_hg: AnalysisReferences,
 ):
     """Test analysis references retrieval to be used for Balsamic human genome analyses."""
 
@@ -602,7 +605,7 @@ def test_get_analysis_references_hg(
 
     # THEN the retrieved analysis references should match the mocked one
     assert type(analysis_references) is AnalysisReferencesHg
-    assert analysis_references.dict() == analysis_references_hg_data
+    assert analysis_references == analysis_references_hg
 
 
 def test_get_analysis_references_canfam(
@@ -620,4 +623,4 @@ def test_get_analysis_references_canfam(
 
     # THEN the retrieved analysis references should match the mocked one
     assert type(analysis_references) is AnalysisReferencesCanFam
-    assert analysis_references.dict() == analysis_references_data
+    assert analysis_references.model_dump() == analysis_references_data