Fix Pacbio read calculation

cg/services/run_devices/pacbio/data_storage_service/pacbio_store_service.py

@@ -56,13 +56,12 @@ def _update_sample(
         sequencing_date: datetime,
     ) -> None:
         """Update the reads and last sequenced date for the SMRT cell samples."""
-        for sample_dto in sample_run_metrics_dtos:
-            self.store.update_sample_reads(
-                internal_id=sample_dto.sample_internal_id, reads=sample_dto.hifi_reads
-            )
-            self.store.update_sample_sequenced_at(
-                internal_id=sample_dto.sample_internal_id, date=sequencing_date
-            )
+        sample_ids_to_update: set[str] = {
+            sample_dto.sample_internal_id for sample_dto in sample_run_metrics_dtos
+        }
+        for sample_id in sample_ids_to_update:
+            self.store.recalculate_sample_reads_pacbio(sample_id)
+            self.store.update_sample_sequenced_at(internal_id=sample_id, date=sequencing_date)
 
     @handle_post_processing_errors(
         to_except=(PostProcessingDataTransferError, ValueError),

cg/store/crud/create.py

@@ -552,7 +552,7 @@ def create_pac_bio_sample_sequencing_run(
     ) -> PacbioSampleSequencingMetrics:
         sample_id: str = sample_run_metrics_dto.sample_internal_id
         LOG.debug(f"Creating Pacbio sample sequencing metric for sample {sample_id}")
-        sample: Sample = self.get_sample_by_internal_id(sample_id)
+        sample: Sample = self.get_sample_by_internal_id_strict(sample_id)
         if not sample:
             self.rollback()
             raise EntryNotFoundError(f"Sample not found: {sample_id}")

cg/store/crud/update.py

@@ -78,10 +78,10 @@ def update_sample_reads_illumina(self, internal_id: str, sequencer_type: Sequenc
         sample.reads = total_reads_for_sample
         self.commit_to_store()
 
-    def update_sample_reads(self, internal_id: str, reads: int):
+    def update_sample_reads_pacbio(self, internal_id: str, reads: int):
         """Add reads to the current reads for a sample."""
-        sample: Sample = self.get_sample_by_internal_id(internal_id)
-        sample.reads += reads
+        sample: Sample = self.get_sample_by_internal_id_strict(internal_id)
+        sample.reads = reads
         self.commit_to_store()
 
     def update_sample_sequenced_at(self, internal_id: str, date: datetime):

cg/store/store.py

@@ -10,4 +10,11 @@ class Store(
     DeleteMixin,
     UpdateMixin,
 ):
-    pass
+    def recalculate_sample_reads_pacbio(self, sample_id: str) -> None:
+        reads = sum(
+            metric.hifi_reads
+            for metric in self.get_pacbio_sample_sequencing_metrics(
+                sample_id=sample_id, smrt_cell_ids=None
+            )
+        )
+        self.update_sample_reads_pacbio(internal_id=sample_id, reads=reads)

tests/store/api/test_base.py

@@ -2,9 +2,15 @@
 
 from sqlalchemy.orm import Query
 
+from cg.constants.devices import DeviceType
 from cg.constants.subject import PhenotypeStatus
-from cg.store.models import CaseSample
+from cg.services.run_devices.pacbio.data_transfer_service.dto import (
+    PacBioSampleSequencingMetricsDTO,
+    PacBioSMRTCellDTO,
+)
+from cg.store.models import CaseSample, PacbioSequencingRun
 from cg.store.store import Store
+from tests.store_helpers import StoreHelpers
 
 
 def test_get_latest_analyses_for_cases_query(
@@ -45,3 +51,50 @@ def test_get_latest_analyses_for_cases_query(
     # THEN only the newest analysis should be returned
     assert analysis_newest in analyses
     assert analysis_oldest not in analyses
+
+
+def test_update_pacbio_sample_reads(base_store: Store, helpers: StoreHelpers):
+
+    # GIVEN a sample in the database with some initial reads and corresponding sample sequencing metrics
+    initial_reads = 1000
+    sample_id = "sample_id"
+    helpers.add_sample(store=base_store, internal_id=sample_id, reads=initial_reads)
+    pacbio_smrt_cell = PacBioSMRTCellDTO(type=DeviceType.PACBIO, internal_id="EA123")
+    pacbio_smrt_cell = base_store.create_pac_bio_smrt_cell(pacbio_smrt_cell)
+    base_store.commit_to_store()
+    sequencing_run: PacbioSequencingRun = helpers.add_pacbio_sequencing_run(
+        store=base_store, id=1, device_id=pacbio_smrt_cell.id, run_name="run_name"
+    )
+    sample_run_metrics_dto = PacBioSampleSequencingMetricsDTO(
+        sample_internal_id=sample_id,
+        hifi_reads=initial_reads,
+        hifi_yield=1,
+        hifi_mean_read_length=1,
+        hifi_median_read_quality="good",
+        polymerase_mean_read_length=1,
+    )
+    base_store.create_pac_bio_sample_sequencing_run(
+        sample_run_metrics_dto=sample_run_metrics_dto, sequencing_run=sequencing_run
+    )
+    base_store.commit_to_store()
+
+    # GIVEN that there are additional sample sequencing metrics for the sample
+    new_reads = 2000
+    new_sample_run_metrics_dto = PacBioSampleSequencingMetricsDTO(
+        sample_internal_id=sample_id,
+        hifi_reads=new_reads,
+        hifi_yield=1,
+        hifi_mean_read_length=1,
+        hifi_median_read_quality="good",
+        polymerase_mean_read_length=1,
+    )
+    base_store.create_pac_bio_sample_sequencing_run(
+        sample_run_metrics_dto=new_sample_run_metrics_dto, sequencing_run=sequencing_run
+    )
+    base_store.commit_to_store()
+
+    # WHEN updating a samples reads
+    base_store.recalculate_sample_reads_pacbio("sample_id")
+
+    # THEN the sample should have updated the reads to the sum of the sample sequencing metrics
+    assert base_store.get_sample_by_internal_id_strict(sample_id).reads == initial_reads + new_reads

tests/store/crud/update/test_update.py

@@ -123,13 +123,34 @@ def test_update_sample_reads_pacbio(
     reads: int = 10000
 
     # WHEN updating the reads for the sample
-    store.update_sample_reads(internal_id=pacbio_barcoded_sample_internal_id, reads=reads)
+    store.update_sample_reads_pacbio(internal_id=pacbio_barcoded_sample_internal_id, reads=reads)
 
     # THEN the reads for the sample is updated
     sample: Sample = store.get_sample_by_internal_id(pacbio_barcoded_sample_internal_id)
     assert sample.reads == reads
 
 
+def test_update_sample_reads_pacbio_not_incremented(
+    pacbio_barcoded_sample_internal_id: str,
+    store: Store,
+    helpers: StoreHelpers,
+):
+    """Tests that updating the reads for a PacBio sample does not increment the reads."""
+    # GIVEN a store with a PacBio sample with reads
+    sample: Sample = helpers.add_sample(store=store, internal_id=pacbio_barcoded_sample_internal_id)
+    sample.reads = 1
+    new_reads = 10000
+
+    # WHEN updating the reads for the sample
+    store.update_sample_reads_pacbio(
+        internal_id=pacbio_barcoded_sample_internal_id, reads=new_reads
+    )
+
+    # THEN the reads for the sample are updated
+    sample: Sample = store.get_sample_by_internal_id_strict(pacbio_barcoded_sample_internal_id)
+    assert sample.reads == new_reads
+
+
 @pytest.mark.parametrize(
     "sample_id_fixture",
     ["sample_id_sequenced_on_multiple_flow_cells", "pacbio_barcoded_sample_internal_id"],