update to unittests; in-work effort

pyproject.toml

@@ -25,7 +25,7 @@ pandas = "^2.0.3"
 snakemake = "^7.32.3"
 pysam = "^0.21.0"
 ryd = "^0.7.1"
-pydantic = "^2.0.0"
+pydantic = "^1.10.12"
 networkx = "^3.1"
 Jinja2 = "^3.1.2"
 more-itertools = "^10.1.0"

src/cgr_gwas_qc/cli/pre_flight.py

@@ -125,7 +125,7 @@ def main(
 def check_config(filename: Path) -> Config:
     try:
         data = yaml.load(filename)
-        config = Config.model_validate(data)
+        config = Config.parse_obj(data)
     except Exception as err:
         if isinstance(err, OSError):
             msg = err.args[1]

src/cgr_gwas_qc/config.py

@@ -64,7 +64,7 @@ def __init__(self, root: Path, user_config: Path, sample_sheet_file: Path):
         self.sample_sheet_file: Path = sample_sheet_file
 
         data = yaml.load(self.user_config)
-        self._config = Config.model_validate(data)
+        self._config = Config.parse_obj(data)
         self._sample_sheet: pd.DataFrame = sample_sheet.read(self.sample_sheet_file)
         self._cluster_groups = sorted(self.ss.cluster_group.unique())
 
@@ -244,4 +244,4 @@ def serialize(data):
                 cleaned[k] = v
         return cleaned
 
-    yaml.write(serialize(cfg.model_dump(exclude_none=exclude_none, **kwargs)), Path(yaml_file))
+    yaml.write(serialize(cfg.dict(exclude_none=exclude_none, **kwargs)), Path(yaml_file))

src/cgr_gwas_qc/models/config/user_files.py

@@ -105,7 +105,7 @@ def validate_gtc_pattern(cls, v):
 
         return v
 
-    @root_validator(skip_on_failure=True)
+    @root_validator
     def check_ped_map(cls, values):
         ped, map_ = values.get("ped"), values.get("map")
         if ped is not None and map_ is None:
@@ -116,7 +116,7 @@ def check_ped_map(cls, values):
 
         return values
 
-    @root_validator(skip_on_failure=True)
+    @root_validator
     def check_bed_bim_fam(cls, values):
         bed, bim, fam = values.get("bed"), values.get("bim"), values.get("fam")
 

src/cgr_gwas_qc/testing/data.py

@@ -138,7 +138,7 @@ def make_cgr_sample_sheet(self, **kwargs) -> U:
 
         Raises:
             ValueError: You must have provided a working directory when
-              instantiating the object.
+                        instantiating the object.
 
         Returns:
             Creates a file ``working_dir/cgr_sample_sheet.csv`` and returns
@@ -169,7 +169,7 @@ def add_user_files(self, entry_point: str = "bed", copy: bool = True) -> U:
 
         Args:
             entry_point: Which entry point to use ("gtc", "ped", or "bed").
-              Defaults to "bed".
+                            Defaults to "bed".
 
         """
         if self.working_dir and copy:

tests/cli/test_pre_flight.py

@@ -88,7 +88,7 @@ def test_check_reference_files_bpm_error(fake_cfg: ConfigMgr, capsys):
     # GIVEN: a working directory
     with chdir(fake_cfg.root):
         # WHEN: Run `cgr pre-flight`
-        reference_files = fake_cfg.config.reference_files.model_copy()
+        reference_files = fake_cfg.config.reference_files.copy()
         reference_files.illumina_manifest_file = "missing.bpm"
         pre_flight.check_reference_files(reference_files)
         # THEN: No errors and `cgr_sample_sheet.csv` exists
@@ -100,7 +100,7 @@ def test_check_reference_files_vcf_error(fake_cfg: ConfigMgr, capsys):
     # GIVEN: a working directory
     with chdir(fake_cfg.root):
         # WHEN: Run `cgr pre-flight`
-        reference_files = fake_cfg.config.reference_files.model_copy()
+        reference_files = fake_cfg.config.reference_files.copy()
         reference_files.thousand_genome_vcf = "missing.vfc"
         pre_flight.check_reference_files(reference_files)
         # THEN: No errors and `cgr_sample_sheet.csv` exists
@@ -112,7 +112,7 @@ def test_check_reference_files_tbi_error(fake_cfg: ConfigMgr, capsys):
     # GIVEN: a working directory
     with chdir(fake_cfg.root):
         # WHEN: Run `cgr pre-flight`
-        reference_files = fake_cfg.config.reference_files.model_copy()
+        reference_files = fake_cfg.config.reference_files.copy()
         reference_files.thousand_genome_tbi = "missing.vfc.tbi"
         pre_flight.check_reference_files(reference_files)
         # THEN: No errors and `cgr_sample_sheet.csv` exists

tests/cluster_profiles/test_cgems.py

@@ -9,6 +9,10 @@
     os.environ.get("GITHUB_ACTIONS", "false") == "true", reason="Broken on GitHub Actions"
 )
 def test_cgems_submission_end_to_end(tmp_path, qsub):
+    # tmp_path = "src/cgr_gwas_qc/testing"
+    print("!!!!!!!!!")
+    print(tmp_path)
+    print("!!!!!!!!!")
     with pytest.raises(SystemExit) as exc:
         args = [
             "-j",

tests/workflow/scripts/test_sample_concordance.py

@@ -55,12 +55,14 @@ def test_add_expected_replicate_missing_pair(fake_cfg, concordance):
     # GIVEN: That I have an expected replicate that is not in the concordance table
     ss = fake_cfg.ss.copy()
     ss.loc[0, "replicate_ids"] = "SP00001|SP10000"  # Add an expected replicate
-    ss = ss.append(
-        pd.Series({"Sample_ID": "SP10000", "replicate_ids": "SP00001|SP10000"}), ignore_index=True
-    )
+    pseudo_data = pd.DataFrame({"Sample_ID": ["SP10000"], "replicate_ids": ["SP00001|SP10000"]})
+    ss = pd.concat([ss, pseudo_data], ignore_index=True)
 
     # THEN: I should add that pair and flag them as an expected_replicate
     df = concordance.pipe(sample_concordance._add_expected_replicates, ss)
+    print("!!!!!!!!")
+    print(df["is_expected_replicate"].values)
+    print("!!!!!!!!")
     assert 2 == df.is_expected_replicate.sum()