add new mutant validation rule (#4075)

## Description
Closes Clinical-Genomics/improve-order-flow#102
Fixes the sample name validation rule for Mutant orders to exclude control samples from the name check

### Added
- Validation function `validate_non_control_sample_names_available` that verifies that sample names used in an order are not in the store excluding control samples
- Utils function `get_sample_name_not_available_errors` used in new validation function
- Mutant order fixture
- aliases to group samples with organism and samples orders with control samples
- tests for new validation functions

### Changed
- Refactored validation function `validate_sample_names_available` to use the new utils function

### Fixed

- Changed some balsamic structure in the order validation service
- Reverted some deletions in json order fixtures from previous prs

cg/services/order_validation_service/errors/sample_errors.py

@@ -61,6 +61,11 @@ class SampleNameNotAvailableError(SampleError):
     message: str = "Sample name already used in previous order"
 
 
+class SampleNameNotAvailableControlError(SampleError):
+    field: str = "name"
+    message: str = "Sample name already in use. Only control samples are allowed repeated names"
+
+
 class ContainerNameRepeatedError(SampleError):
     field: str = "container_name"
     message: str = "Tube names must be unique among samples"

cg/services/order_validation_service/models/order_aliases.py

@@ -1,7 +1,17 @@
 from cg.services.order_validation_service.workflows.fluffy.models.order import FluffyOrder
+from cg.services.order_validation_service.workflows.metagenome.models.sample import MetagenomeSample
+from cg.services.order_validation_service.workflows.microbial_fastq.models.order import (
+    MicrobialFastqOrder,
+)
 from cg.services.order_validation_service.workflows.microsalt.models.order import MicrosaltOrder
 from cg.services.order_validation_service.workflows.mutant.models.order import MutantOrder
 from cg.services.order_validation_service.workflows.rml.models.order import RmlOrder
+from cg.services.order_validation_service.workflows.taxprofiler.models.sample import (
+    TaxprofilerSample,
+)
 
 OrderWithIndexedSamples = FluffyOrder | RmlOrder
-OrderWithNonHumanSamples = MutantOrder | MicrosaltOrder
+OrderWithSamplesFromOrganism = MutantOrder | MicrosaltOrder
+OrderWithControlSamples = (
+    MetagenomeSample | MicrobialFastqOrder | MicrosaltOrder | MutantOrder | TaxprofilerSample
+)

cg/services/order_validation_service/models/sample_aliases.py

@@ -4,18 +4,15 @@
 )
 from cg.services.order_validation_service.workflows.fastq.models.sample import FastqSample
 from cg.services.order_validation_service.workflows.fluffy.models.sample import FluffySample
-from cg.services.order_validation_service.workflows.microsalt.models.sample import MicrosaltSample
 from cg.services.order_validation_service.workflows.mip_dna.models.sample import MipDnaSample
 from cg.services.order_validation_service.workflows.mip_rna.models.sample import MipRnaSample
-from cg.services.order_validation_service.workflows.mutant.models.sample import MutantSample
 from cg.services.order_validation_service.workflows.rml.models.sample import RmlSample
 from cg.services.order_validation_service.workflows.rna_fusion.models.sample import RnaFusionSample
 from cg.services.order_validation_service.workflows.tomte.models.sample import TomteSample
 
 HumanSample = (
     BalsamicSample | BalsamicUmiSample | FastqSample | MipDnaSample | RnaFusionSample | TomteSample
 )
-NonHumanSample = MutantSample | MicrosaltSample
 
 IndexedSample = FluffySample | RmlSample
 

cg/services/order_validation_service/rules/sample/rules.py

@@ -1,4 +1,5 @@
 from cg.models.orders.constants import OrderType
+from cg.models.orders.sample_base import ControlEnum
 from cg.services.order_validation_service.errors.sample_errors import (
     ApplicationArchivedError,
     ApplicationNotCompatibleError,
@@ -22,14 +23,16 @@
     WellPositionRmlMissingError,
 )
 from cg.services.order_validation_service.models.order_aliases import (
+    OrderWithControlSamples,
     OrderWithIndexedSamples,
-    OrderWithNonHumanSamples,
+    OrderWithSamplesFromOrganism,
 )
 from cg.services.order_validation_service.models.sample_aliases import IndexedSample
 from cg.services.order_validation_service.rules.sample.utils import (
     PlateSamplesValidator,
     get_indices_for_repeated_sample_names,
     get_indices_for_tube_repeated_container_name,
+    get_sample_name_not_available_errors,
     has_multiple_applications,
     has_multiple_priorities,
     is_container_name_missing,
@@ -155,11 +158,11 @@ def validate_concentration_required_if_skip_rc(
 
 
 def validate_organism_exists(
-    order: OrderWithNonHumanSamples, store: Store, **kwargs
+    order: OrderWithSamplesFromOrganism, store: Store, **kwargs
 ) -> list[OrganismDoesNotExistError]:
     """
     Validate that the organisms of all samples in the order exist in the database.
-    Only applicable to order types with non-human samples.
+    Only applicable to Microsalt and Mutant orders.
     """
     errors: list[OrganismDoesNotExistError] = []
     for sample_index, sample in order.enumerated_samples:
@@ -208,16 +211,24 @@ def validate_sample_names_available(
 ) -> list[SampleNameNotAvailableError]:
     """
     Validate that the sample names do not exists in the database under the same customer.
-    Applicable to all order types.
+    Applicable to all orders without control samples.
     """
-    errors: list[SampleNameNotAvailableError] = []
-    customer = store.get_customer_by_internal_id(order.customer)
-    for sample_index, sample in order.enumerated_samples:
-        if store.get_sample_by_customer_and_name(
-            sample_name=sample.name, customer_entry_id=[customer.id]
-        ):
-            error = SampleNameNotAvailableError(sample_index=sample_index)
-            errors.append(error)
+    errors: list[SampleNameNotAvailableError] = get_sample_name_not_available_errors(
+        order=order, store=store, has_order_control=False
+    )
+    return errors
+
+
+def validate_non_control_sample_names_available(
+    order: OrderWithControlSamples, store: Store, **kwargs
+) -> list[SampleNameNotAvailableError]:
+    """
+    Validate that non-control sample names do not exists in the database under the same customer.
+    Applicable to all orders with control samples.
+    """
+    errors: list[SampleNameNotAvailableError] = get_sample_name_not_available_errors(
+        order=order, store=store, has_order_control=True
+    )
     return errors
 
 
@@ -226,7 +237,7 @@ def validate_sample_names_unique(
 ) -> list[SampleNameRepeatedError]:
     """
     Validate that all the sample names are unique within the order.
-    Applicable to all order types except Mutant orders.
+    Applicable to all order types.
     """
     sample_indices: list[int] = get_indices_for_repeated_sample_names(order)
     return [SampleNameRepeatedError(sample_index=sample_index) for sample_index in sample_indices]

cg/services/order_validation_service/rules/sample/utils.py

@@ -1,13 +1,16 @@
 import re
 from collections import Counter
 
-from cg.models.orders.sample_base import ContainerEnum
+from cg.models.orders.sample_base import ContainerEnum, ControlEnum
 from cg.services.order_validation_service.constants import ALLOWED_SKIP_RC_BUFFERS
 from cg.services.order_validation_service.errors.sample_errors import (
     BufferInvalidError,
     ConcentrationInvalidIfSkipRCError,
     ConcentrationRequiredError,
     OccupiedWellError,
+    SampleError,
+    SampleNameNotAvailableControlError,
+    SampleNameNotAvailableError,
     WellPositionMissingError,
 )
 from cg.services.order_validation_service.models.order_with_samples import OrderWithSamples
@@ -79,6 +82,45 @@ def get_indices_for_repeated_sample_names(order: OrderWithSamples) -> list[int]:
     return indices
 
 
+def get_sample_name_not_available_errors(
+    order: OrderWithSamples, store: Store, has_order_control: bool
+) -> list[SampleError]:
+    """Return errors for non-control samples with names already used in the database."""
+    errors: list[SampleError] = []
+    customer = store.get_customer_by_internal_id(order.customer)
+    for sample_index, sample in order.enumerated_samples:
+        if store.get_sample_by_customer_and_name(
+            sample_name=sample.name, customer_entry_id=[customer.id]
+        ):
+            if is_sample_name_allowed_to_be_repeated(has_control=has_order_control, sample=sample):
+                continue
+            error = get_appropriate_sample_name_available_error(
+                has_control=has_order_control, sample_index=sample_index
+            )
+            errors.append(error)
+    return errors
+
+
+def is_sample_name_allowed_to_be_repeated(has_control: bool, sample: Sample) -> bool:
+    """
+    Return whether a sample name can be used if it is already in the database.
+    This is the case when the order has control samples and the sample is a control.
+    """
+    return has_control and sample.control in [ControlEnum.positive, ControlEnum.negative]
+
+
+def get_appropriate_sample_name_available_error(
+    has_control: bool, sample_index: int
+) -> SampleError:
+    """
+    Return the appropriate error for a sample name that is not available based on whether the
+    order has control samples or not.
+    """
+    if has_control:
+        return SampleNameNotAvailableControlError(sample_index=sample_index)
+    return SampleNameNotAvailableError(sample_index=sample_index)
+
+
 def is_tube_container_name_redundant(sample: Sample, counter: Counter) -> bool:
     return sample.container == ContainerEnum.tube and counter.get(sample.container_name) > 1
 

cg/services/order_validation_service/workflows/balsamic/rules/case_sample/rules.py → cg/services/order_validation_service/workflows/balsamic/rules.py

@@ -1,6 +1,6 @@
 from cg.services.order_validation_service.errors.case_sample_errors import CaptureKitMissingError
 from cg.services.order_validation_service.workflows.balsamic.models.order import BalsamicOrder
-from cg.services.order_validation_service.workflows.balsamic.rules.case_sample.utils import (
+from cg.services.order_validation_service.workflows.balsamic.utils import (
     is_sample_missing_capture_kit,
 )
 from cg.store.store import Store

cg/services/order_validation_service/workflows/balsamic/validation_rules.py

@@ -25,7 +25,7 @@
     validate_well_positions_required,
     validate_wells_contain_at_most_one_sample,
 )
-from cg.services.order_validation_service.workflows.balsamic.rules.case_sample.rules import (
+from cg.services.order_validation_service.workflows.balsamic.rules import (
     validate_capture_kit_panel_requirement,
 )
 

cg/services/order_validation_service/workflows/metagenome/validation_rules.py

@@ -3,7 +3,7 @@
     validate_application_exists,
     validate_applications_not_archived,
     validate_container_name_required,
-    validate_sample_names_available,
+    validate_non_control_sample_names_available,
     validate_sample_names_unique,
     validate_tube_container_name_unique,
     validate_volume_interval,
@@ -18,7 +18,7 @@
     validate_application_exists,
     validate_applications_not_archived,
     validate_container_name_required,
-    validate_sample_names_available,
+    validate_non_control_sample_names_available,
     validate_sample_names_unique,
     validate_tube_container_name_unique,
     validate_volume_interval,

cg/services/order_validation_service/workflows/microbial_fastq/validation_rules.py

@@ -3,7 +3,7 @@
     validate_application_exists,
     validate_applications_not_archived,
     validate_container_name_required,
-    validate_sample_names_available,
+    validate_non_control_sample_names_available,
     validate_sample_names_unique,
     validate_tube_container_name_unique,
     validate_volume_interval,
@@ -18,7 +18,7 @@
     validate_application_exists,
     validate_applications_not_archived,
     validate_container_name_required,
-    validate_sample_names_available,
+    validate_non_control_sample_names_available,
     validate_sample_names_unique,
     validate_tube_container_name_unique,
     validate_volume_interval,

cg/services/order_validation_service/workflows/microsalt/validation_rules.py

@@ -3,8 +3,8 @@
     validate_application_exists,
     validate_applications_not_archived,
     validate_container_name_required,
+    validate_non_control_sample_names_available,
     validate_organism_exists,
-    validate_sample_names_available,
     validate_sample_names_unique,
     validate_tube_container_name_unique,
     validate_volume_interval,
@@ -19,8 +19,8 @@
     validate_application_exists,
     validate_applications_not_archived,
     validate_container_name_required,
+    validate_non_control_sample_names_available,
     validate_organism_exists,
-    validate_sample_names_available,
     validate_sample_names_unique,
     validate_tube_container_name_unique,
     validate_volume_interval,

cg/services/order_validation_service/workflows/mutant/validation_rules.py

@@ -3,8 +3,8 @@
     validate_application_exists,
     validate_applications_not_archived,
     validate_container_name_required,
+    validate_non_control_sample_names_available,
     validate_organism_exists,
-    validate_sample_names_available,
     validate_sample_names_unique,
     validate_tube_container_name_unique,
     validate_volume_interval,
@@ -19,9 +19,9 @@
     validate_application_exists,
     validate_applications_not_archived,
     validate_container_name_required,
+    validate_non_control_sample_names_available,
     validate_organism_exists,
     validate_volume_required,
-    validate_sample_names_available,
     validate_sample_names_unique,
     validate_tube_container_name_unique,
     validate_volume_interval,

cg/services/order_validation_service/workflows/taxprofiler/validation_rules.py

@@ -3,7 +3,7 @@
     validate_application_exists,
     validate_applications_not_archived,
     validate_container_name_required,
-    validate_sample_names_available,
+    validate_non_control_sample_names_available,
     validate_sample_names_unique,
     validate_tube_container_name_unique,
     validate_volume_interval,
@@ -18,7 +18,7 @@
     validate_application_exists,
     validate_applications_not_archived,
     validate_container_name_required,
-    validate_sample_names_available,
+    validate_non_control_sample_names_available,
     validate_sample_names_unique,
     validate_tube_container_name_unique,
     validate_volume_interval,

tests/fixture_plugins/orders_fixtures/order_fixtures.py

@@ -12,6 +12,7 @@
 from cg.services.order_validation_service.workflows.microsalt.models.order import MicrosaltOrder
 from cg.services.order_validation_service.workflows.mip_dna.models.order import MipDnaOrder
 from cg.services.order_validation_service.workflows.mip_rna.models.order import MipRnaOrder
+from cg.services.order_validation_service.workflows.mutant.models.order import MutantOrder
 from cg.services.order_validation_service.workflows.pacbio_long_read.models.order import PacbioOrder
 from cg.services.order_validation_service.workflows.rml.models.order import RmlOrder
 
@@ -88,6 +89,14 @@ def mip_rna_order(mip_rna_order_to_submit: dict) -> MipRnaOrder:
     return mip_rna_order
 
 
+@pytest.fixture
+def mutant_order(sarscov2_order_to_submit: dict, ticket_id_as_int: int) -> MutantOrder:
+    """Parse mutant order example."""
+    order = MutantOrder.model_validate(sarscov2_order_to_submit)
+    order._generated_ticket_id = ticket_id_as_int
+    return order
+
+
 @pytest.fixture
 def pacbio_order(pacbio_order_to_submit: dict, ticket_id_as_int: int) -> PacbioOrder:
     order = PacbioOrder.model_validate(pacbio_order_to_submit)