Replace hardcoded validation cases

alembic/versions/2023_10_31_fce8a2ca0fd1_add_is_compressible.py

@@ -0,0 +1,26 @@
+"""Add is_compressible
+
+Revision ID: fce8a2ca0fd1
+Revises: 9073c61bc72b
+Create Date: 2023-10-31 10:18:11.450637
+
+"""
+from alembic import op
+import sqlalchemy as sa
+
+# revision identifiers, used by Alembic.
+revision = "fce8a2ca0fd1"
+down_revision = "9073c61bc72b"
+branch_labels = None
+depends_on = None
+
+
+def upgrade():
+    op.add_column(
+        "family",
+        sa.Column("is_compressible", sa.Boolean(), nullable=False, server_default=sa.text("1")),
+    )
+
+
+def downgrade():
+    op.drop_column("family", "is_compressible")

cg/cli/compress/fastq.py

@@ -10,7 +10,6 @@
     compress_sample_fastqs_in_cases,
     correct_spring_paths,
     get_cases_to_process,
-    is_case_ignored,
     update_compress_api,
 )
 from cg.constants.constants import DRY_RUN
@@ -91,8 +90,6 @@ def clean_fastq(context: CGConfig, case_id: Optional[str], days_back: int, dry_r
 
     cleaned_inds = 0
     for case in cases:
-        if is_case_ignored(case_id=case.internal_id):
-            continue
         samples: Iterable[str] = store.get_sample_ids_by_case_id(case_id=case.internal_id)
         for sample_id in samples:
             was_cleaned: bool = compress_api.clean_fastq(sample_id=sample_id)

cg/cli/compress/helpers.py

@@ -10,7 +10,6 @@
 
 from cg.apps.housekeeper.hk import HousekeeperAPI
 from cg.constants.compression import (
-    CASES_TO_IGNORE,
     CRUNCHY_MIN_GB_PER_PROCESS,
     MAX_READS_PER_GB,
 )
@@ -31,11 +30,12 @@ def get_cases_to_process(
     """Return cases to process."""
     cases: list[Family] = []
     if case_id:
-        case: Family = store.get_case_by_internal_id(internal_id=case_id)
+        case: Family = store.get_case_by_internal_id(case_id)
         if not case:
             LOG.warning(f"Could not find case {case_id}")
             return
-        cases.append(case)
+        if case.is_compressible:
+            cases.append(case)
     else:
         date_threshold: dt.datetime = get_date_days_ago(days_ago=days_back)
         cases: list[Family] = store.get_cases_to_compress(date_threshold=date_threshold)
@@ -53,14 +53,6 @@ def get_fastq_individuals(store: Store, case_id: str = None) -> Iterator[str]:
         yield link_obj.sample.internal_id
 
 
-def is_case_ignored(case_id: str) -> bool:
-    """Check if case should be skipped."""
-    if case_id in CASES_TO_IGNORE:
-        LOG.debug(f"Skipping case: {case_id}")
-        return True
-    return False
-
-
 def set_memory_according_to_reads(
     sample_id: str, sample_reads: Optional[int] = None, sample_process_mem: Optional[int] = None
 ) -> Optional[int]:
@@ -150,8 +142,6 @@ def compress_sample_fastqs_in_cases(
         case_converted = True
         if case_conversion_count >= number_of_conversions:
             break
-        if is_case_ignored(case_id=case.internal_id):
-            continue
 
         LOG.info(f"Searching for FASTQ files in case {case.internal_id}")
         if not case.links:

cg/constants/compression.py

@@ -13,168 +13,3 @@
 # Number of days until FASTQs counts as old
 FASTQ_DELTA = 21
 FASTQ_DATETIME_DELTA = datetime.timedelta(days=FASTQ_DELTA)
-
-PROBLEMATIC_CASES = [
-    "causalmite",
-    "deepcub",
-    "expertalien",
-    "fluenteagle",
-    "grandkoi",
-    "lovingmayfly",
-    "loyalegret",
-    "modernbee",
-    "proudcollie",
-    "richalien",
-    "suremako",
-    "wisestork",
-]
-
-# List of cases used for validation that we should skip
-BALSAMIC_VALIDATION_CASES = [
-    "setamoeba",  # BALSAMIC validation case tumor-only panel
-    "sweetelf",  # BALSAMIC positive control tumor-only panel
-    "poeticghoul",  # BALSAMIC positive control tumor-only panel
-    "equalbug",  # UMI seracare validation case (AF 0.5%) tumor-normal panel
-    "stableraven",  # UMI seracare validation case (AF 1%) tumor-normal panel
-    "uphippo",  # UMI seracare validation case (AF 0.5%) tumor-only panel
-    "cleanfowl",  # BALSAMIC validation case, HD829 reference for FLT3 Ascertation
-    "proudsquid",  # BALSAMIC validation case, HD829 reference for FLT3 Ascertation
-    "modestjaguar",  # BALSAMIC validation case, HD829 reference for FLT3 Ascertation
-    "dearmarmot",  # BALSAMIC validation case, HD829 reference for FLT3 Ascertation
-    "holykid",  # BALSAMIC validation case, HD829 reference for FLT3 Ascertation
-    "civilsole",  # BALSAMIC validation case tumor-only wgs
-    "fleetjay",  # BALSAMIC validation case tumor-normal wgs
-    "grandmarmot",  # BALSAMIC validation case tumor-normal wgs
-    "unitedbeagle",  # BALSAMIC validation case tumor-normal panel
-    "rightthrush",  # BALSAMIC validation case tumor-only wes
-    "properpigeon",  # BALSAMIC validation case tumor-normal wes
-    "eagerox",  # BALSAMIC validation case from cust087, tumor-only panel
-    "casualweasel",  # BALSAMIC validation case from cust087, tumor-only panel
-    "acetuna",  # BALSAMIC validation case from cust087, tumor-only panel
-    "suitedsnake",  # BALSAMIC validation case from cust087, tumor-only panel
-    "savinghorse",  # BALSAMIC validation case from cust087, tumor-only panel
-    "rightpup",  # BALSAMIC validation case from cust087, tumor-only panel
-    "sureroughy",  # BALSAMIC validation case from cust087, tumor-only panel
-    "notedshark",  # BALSAMIC validation case from cust127, tumor-normal WGS (SV inversion positive control)
-    "wholewhale",  # BALSAMIC validation case from cust143, tumor-normal WGS (SV translocation positive control)
-    "largeturtle",  # BALSAMIC validation case from cust143, tumor-normal WGS (SV translocation positive control)
-    "lightkodiak",  # BALSAMIC validation case from cust143, tumor-normal WGS (SV inversion positive control)
-    "wholecivet",  # BALSAMIC validation case from cust110, tumor-normal WGS (SV inversion positive control)
-    "upwardstork",  # BALSAMIC validation case from cust110, tumor-normal WGS (SV deletion positive control)
-    "suitedgrub",  # BALSAMIC validation case from cust110, tumor-normal WGS (SV deletion positive control)
-]
-
-FLUFFY_VALIDATION_CASES = [
-    "simplesalmon",  # Chromosome 13, 18, 21 Suspected
-]
-
-MIP_VALIDATION_CASES = [
-    "brightcaiman",  # DNA rare disease positive control
-    "casualgannet",  # DNA rare disease positive control
-    "civilkoala",  # RNA rare disease positive control
-    "cleanshrimp",  # DNA rare disease positive control
-    "drivenmolly",  # RNA rare disease positive control
-    "easybeetle",  # DNA rare disease positive control
-    "epicasp",  # DNA rare disease positive control
-    "expertmole",  # RNA rare disease positive control
-    "finequagga",  # RNA rare disease positive control
-    "firstfawn",  # DNA rare disease positive control
-    "gladthrush",  # DNA rare disease positive control
-    "helpedfilly",  # DNA rare disease positive control
-    "hotskink",  # DNA rare disease positive control
-    "inferret",  # DNA rare disease positive control
-    "intentcorgi",  # DNA rare disease positive control
-    "intentmayfly",  # DNA rare disease positive control
-    "justhusky",  # DNA rare disease positive control
-    "kindcaiman",  # DNA rare disease positive control
-    "lightprawn",  # DNA rare disease positive control
-    "livingox",  # DNA rare disease positive control
-    "newaphid",  # RNA rare disease positive control
-    "nextjackal",  # DNA rare disease positive control
-    "modernmule",  # DNA rare disease positive control
-    "moralcattle",  # RNA rare disease positive control
-    "onemite",  # DNA rare disease positive control
-    "opencow",  # DNA rare disease positive control
-    "proudcougar",  # DNA rare disease positive control
-    "rightmacaw",  # DNA rare disease positive control
-    "safeguinea",  # DNA rare disease positive control
-    "sharpparrot",  # RNA rare disease positive control
-    "sharppigeon",  # DNA rare disease positive control
-    "sharpwhale",  # DNA rare disease positive control
-    "stillant",  # DNA rare disease positive control
-    "smoothboa",  # RNA rare disease positive control
-    "strongbison",  # DNA rare disease positive control
-    "tenderoriole",  # DNA rare disease positive control
-    "topsrhino",  # DNA rare disease positive control
-    "usablemarten",  # DNA rare disease positive control
-    "vitalmouse",  # DNA rare disease positive control
-]
-
-# List of cases used for validation that we should skip
-RNAFUSION_VALIDATION_CASES = [
-    "ableheron",
-    "acecoyote",
-    "ampleray",
-    "bossmink",
-    "cuddlyhen",
-    "daringowl",
-    "expertboar",
-    "finerracer",
-    "growndoe",
-    "guidedfeline",
-    "handyturtle",
-    "hardygrouse",
-    "holyrodent",
-    "honestswine",
-    "inlab",
-    "justburro",
-    "movedmule",
-    "nearbyjoey",
-    "oncrab",
-    "politeglider",
-    "rareosprey",
-    "rightmoray",
-    "stablemoray",
-    "starnewt",
-    "tendergoose",
-    "truemole",
-    "truepigeon",
-    "valuedfowl",
-    "vocallocust",
-    "wantedsawfly",
-]
-
-TAXPROFILER_VALIDATION_CASES: list[str] = [
-    "richurchin",
-]
-
-OTHER_VALIDATION_CASES = [
-    "bigdrum",
-    "busycolt",
-    "daringpony",
-    "frankhusky",
-    "gamedeer",
-    "hotviper",
-    "keencalf",
-    "keenviper",
-    "luckyhog",
-    "maturejay",  # sars-cov-2 case
-    "meetpossum",
-    "mintbaboon",
-    "mintyeti",
-    "proeagle",
-    "propercoral",
-    "pumpedcat",
-    "strongman",
-    "truecoyote",
-]
-
-CASES_TO_IGNORE = (
-    PROBLEMATIC_CASES
-    + OTHER_VALIDATION_CASES
-    + BALSAMIC_VALIDATION_CASES
-    + FLUFFY_VALIDATION_CASES
-    + MIP_VALIDATION_CASES
-    + RNAFUSION_VALIDATION_CASES
-    + TAXPROFILER_VALIDATION_CASES
-)

cg/store/api/status.py

@@ -229,6 +229,7 @@ def get_cases_to_compress(self, date_threshold: datetime) -> list[Family]:
         case_filter_functions: list[CaseFilter] = [
             CaseFilter.FILTER_HAS_INACTIVE_ANALYSIS,
             CaseFilter.FILTER_OLD_BY_CREATION_DATE,
+            CaseFilter.FILTER_IS_COMPRESSIBLE,
         ]
         return apply_case_filter(
             filter_functions=case_filter_functions,

cg/store/filters/status_case_filters.py

@@ -190,6 +190,11 @@ def filter_running_cases(cases: Query, **kwargs) -> Query:
     return cases.filter(Family.action == CaseActions.RUNNING)
 
 
+def filter_compressible_cases(cases: Query, **kwargs) -> Query:
+    """Filter cases which are running."""
+    return cases.filter(Family.is_compressible)
+
+
 def order_cases_by_created_at(cases: Query, **kwargs) -> Query:
     """Order cases by created at."""
     return cases.order_by(Family.created_at.desc())
@@ -256,6 +261,7 @@ class CaseFilter(Enum):
     FILTER_HAS_INACTIVE_ANALYSIS: Callable = filter_inactive_analysis_cases
     FILTER_HAS_SEQUENCE: Callable = filter_cases_has_sequence
     FILTER_IS_RUNNING: Callable = filter_running_cases
+    FILTER_IS_COMPRESSIBLE: Callable = filter_compressible_cases
     FILTER_NEW_BY_ORDER_DATE: Callable = filter_newer_cases_by_order_date
     FILTER_NOT_ANALYSED: Callable = filter_cases_not_analysed
     FILTER_OLD_BY_CREATION_DATE: Callable = filter_older_cases_by_creation_date

cg/store/models.py

@@ -390,6 +390,7 @@ class Family(Model, PriorityMixin):
     data_delivery = Column(types.Enum(*list(DataDelivery)))
     id = Column(types.Integer, primary_key=True)
     internal_id = Column(types.String(32), unique=True, nullable=False)
+    is_compressible = Column(types.Boolean, nullable=False, default=True)
     name = Column(types.String(128), nullable=False)
     ordered_at = Column(types.DateTime, default=dt.datetime.now)
     _panels = Column(types.Text)

tests/cli/compress/test_cli_compress_fastq.py

@@ -71,6 +71,26 @@ def test_get_cases_to_process_when_no_case(
     assert f"Could not find case {case_id_does_not_exist}" in caplog.text
 
 
+def test_incompressible_cases_are_not_processable(
+    helpers: StoreHelpers,
+    populated_compress_context: CGConfig,
+):
+    """Test that cases that are marked as incompressible are not processable."""
+
+    # GIVEN a store with a case that is marked as incompressible
+    status_db: Store = populated_compress_context.status_db
+
+    incompressible_case: Family = helpers.add_case(store=status_db, internal_id="incompressible")
+    incompressible_case.created_at = dt.datetime.now() - dt.timedelta(days=1000)
+    incompressible_case.is_compressible = False
+
+    # WHEN retrieving the processable cases
+    processable_cases: list[Family] = get_cases_to_process(days_back=1, store=status_db)
+
+    # THEN assert that the incompressible case is not processable
+    assert incompressible_case not in processable_cases
+
+
 def test_compress_fastq_cli_no_family(compress_context: CGConfig, cli_runner: CliRunner, caplog):
     """Test to run the compress command with a database without samples,"""
     caplog.set_level(logging.DEBUG)