Add integration test for post processing of flat flow cell demultiplexed with bclconvert

tests/conftest.py

@@ -2754,8 +2754,8 @@ def mock_config(rnafusion_dir: Path, rnafusion_case_id: str) -> None:
     )
 
 
-@pytest.fixture
-def expected_total_reads() -> int:
+@pytest.fixture(name="expected_total_reads", scope="session")
+def fixture_expected_total_reads() -> int:
     return 1_000_000
 
 
@@ -2765,8 +2765,8 @@ def fixture_flow_cell_name() -> str:
     return "HVKJCDRXX"
 
 
-@pytest.fixture
-def store_with_sequencing_metrics(
+@pytest.fixture(name="store_with_sequencing_metrics")
+def fixture_store_with_sequencing_metrics(
     store: Store, sample_id: str, expected_total_reads: int, flow_cell_name: str
 ) -> Generator[Store, None, None]:
     """Return a store with multiple samples with sample lane sequencing metrics."""
@@ -2803,18 +2803,6 @@ def store_with_sequencing_metrics(
     yield store
 
 
-@pytest.fixture
-def flow_cell_name_demultiplexed_with_bcl_convert() -> str:
-    return "HY7FFDRX2"
-
-
-@pytest.fixture
-def flow_cell_directory_name_demultiplexed_with_bcl_convert(
-    flow_cell_name_demultiplexed_with_bcl_convert: str,
-):
-    return f"230504_A00689_0804_B{flow_cell_name_demultiplexed_with_bcl_convert}"
-
-
 @pytest.fixture(name="demultiplexed_flow_cells_tmp_directory")
 def fixture_demultiplexed_flow_cells_tmp_directory(tmp_path) -> Path:
     original_dir = Path(

tests/fixtures/apps/demultiplexing/demultiplexed-runs/230505_A00689_0804_BHY7FFDRX2/Reports/Adapter_Metrics.csv

@@ -0,0 +1,5 @@
+Lane,Sample_ID,Sample_Project,index,index2,ReadNumber,AdapterBases,SampleBases,% Adapter Bases
+1,ACC11927A2,405887,GTCCTTCGGC,CTGTGCATGA,1,0,415032696,0.000
+1,ACC11927A2,405887,GTCCTTCGGC,CTGTGCATGA,2,0,1436651640,0.000
+1,ACC11927A5,405887,GTTTCACGAT,TTTGGCCGAA,1,0,576200560,0.000
+1,ACC11927A5,405887,GTTTCACGAT,TTTGGCCGAA,2,0,1994540400,0.000

tests/fixtures/apps/demultiplexing/demultiplexed-runs/230505_A00689_0804_BHY7FFDRX2/Reports/Demultiplex_Stats.csv

@@ -0,0 +1,6 @@
+Lane,SampleID,Sample_Project,Index,# Reads,# Perfect Index Reads,# One Mismatch Index Reads,# Two Mismatch Index Reads,% Reads,% Perfect Index Reads,% One Mismatch Index Reads,% Two Mismatch Index Reads
+1,ACC11927A2,405887,GTCCTTCGGC-CTGTGCATGA,15962796,15962796,0,0,0.0437,1.0000,0.0000,0.0000
+1,ACC11927A5,405887,GTTTCACGAT-TTTGGCCGAA,22161560,22161560,0,0,0.0607,1.0000,0.0000,0.0000
+1,Undetermined,Undetermined,,38680101,38680101,0,0,0.1059,1.0000,0.0000,0.0000
+2,ACC11927A2,405887,GTCCTTCGGC-CTGTGCATGA,15855013,15855013,0,0,0.0436,1.0000,0.0000,0.0000
+2,ACC11927A5,405887,GTTTCACGAT-TTTGGCCGAA,22146552,22146552,0,0,0.0609,1.0000,0.0000,0.0000

tests/fixtures/apps/demultiplexing/demultiplexed-runs/230505_A00689_0804_BHY7FFDRX2/Reports/Quality_Metrics.csv

@@ -0,0 +1,10 @@
+Lane,SampleID,Sample_Project,index,index2,ReadNumber,Yield,YieldQ30,QualityScoreSum,Mean Quality Score (PF),% Q30
+1,ACC11927A2,405887,GTCCTTCGGC,CTGTGCATGA,1,415032696,393745856,15004333259,36.15,0.95
+1,ACC11927A2,405887,GTCCTTCGGC,CTGTGCATGA,2,1436651640,1344252463,51555254683,35.89,0.94
+1,ACC11927A5,405887,GTTTCACGAT,TTTGGCCGAA,1,576200560,546413091,20826965934,36.15,0.95
+1,ACC11927A5,405887,GTTTCACGAT,TTTGGCCGAA,2,1994540400,1899135721,72196352070,36.20,0.95
+1,Undetermined,Undetermined,,,2,3481209090,2735317061,113977736645,32.74,0.79
+2,ACC11927A2,405887,GTCCTTCGGC,CTGTGCATGA,1,412230338,391068079,14902615287,36.15,0.95
+2,ACC11927A2,405887,GTCCTTCGGC,CTGTGCATGA,2,1426951170,1332820556,51164851641,35.86,0.93
+2,ACC11927A5,405887,GTTTCACGAT,TTTGGCCGAA,1,575810352,545939221,20810867373,36.14,0.95
+2,ACC11927A5,405887,GTTTCACGAT,TTTGGCCGAA,2,1993189680,1896844852,72130263160,36.19,0.95

tests/fixtures/apps/demultiplexing/demultiplexed-runs/230505_A00689_0804_BHY7FFDRX2/SampleSheet.csv

@@ -0,0 +1,9 @@
+[Settings]
+BarcodeMismatchesIndex1,0
+BarcodeMismatchesIndex2,0
+[Data]
+FCID,Lane,Sample_ID,SampleRef,index,index2,SampleName,Control,Recipe,Operator,Sample_Project
+HY7FFDRX2,1,ACC11927A2,hg19,GTCCTTCGGC,CTGTGCATGA,anonymous_1,N,R1,script,405887
+HY7FFDRX2,1,ACC11927A5,hg19,GTTTCACGAT,TTTGGCCGAA,anonymous_2,N,R1,script,405887
+HY7FFDRX2,2,ACC11927A2,hg19,GTCCTTCGGC,CTGTGCATGA,anonymous_1,N,R1,script,405887
+HY7FFDRX2,2,ACC11927A5,hg19,GTTTCACGAT,TTTGGCCGAA,anonymous_2,N,R1,script,405887

tests/meta/demultiplex/conftest.py

@@ -1,8 +1,9 @@
 import pytest
 
+from collections import namedtuple
 from datetime import datetime
 from pathlib import Path
-from typing import List
+from typing import List, Dict
 
 from cg.apps.cgstats.stats import StatsAPI
 from cg.apps.housekeeper.hk import HousekeeperAPI
@@ -14,6 +15,8 @@
 from cg.store.models import Sample, Family
 from tests.store_helpers import StoreHelpers
 
+FlowCellInfo = namedtuple("FlowCellInfo", "directory name sample_internal_ids")
+
 
 @pytest.fixture(name="tmp_demulitplexing_dir")
 def fixture_tmp_demulitplexing_dir(
@@ -339,6 +342,59 @@ def fixture_delete_demultiplex_api(
     )
 
 
+@pytest.fixture(name="flow_cell_info_map", scope="session")
+def fixture_flow_cell_info_map(
+    flow_cell_directory_name_demultiplexed_with_bcl_convert_flat,
+    flow_cell_name_demultiplexed_with_bcl_convert,
+    flow_cell_name_demultiplexed_with_bcl2fastq,
+    bcl_convert_demultiplexed_flow_cell_sample_internal_ids,
+    bcl2fastq_demultiplexed_flow_cell_sample_internal_ids,
+    flow_cell_directory_name_demultiplexed_with_bcl_convert,
+    flow_cell_directory_name_demultiplexed_with_bcl2fastq,
+) -> Dict[str, FlowCellInfo]:
+    """Returns a dict with the suitable fixtures for different demultiplexing softwares and
+    settings. Keys are string, values are named tuples FlowCellInfo."""
+    return {
+        "BCL2FASTQ_TREE": FlowCellInfo(
+            directory=flow_cell_directory_name_demultiplexed_with_bcl2fastq,
+            name=flow_cell_name_demultiplexed_with_bcl2fastq,
+            sample_internal_ids=bcl2fastq_demultiplexed_flow_cell_sample_internal_ids,
+        ),
+        "BCLCONVERT_FLAT": FlowCellInfo(
+            directory=flow_cell_directory_name_demultiplexed_with_bcl_convert_flat,
+            name=flow_cell_name_demultiplexed_with_bcl_convert,
+            sample_internal_ids=bcl_convert_demultiplexed_flow_cell_sample_internal_ids,
+        ),
+        "BCLCONVERT_TREE": FlowCellInfo(
+            directory=flow_cell_directory_name_demultiplexed_with_bcl_convert,
+            name=flow_cell_name_demultiplexed_with_bcl_convert,
+            sample_internal_ids=bcl_convert_demultiplexed_flow_cell_sample_internal_ids,
+        ),
+    }
+
+
+@pytest.fixture(name="flow_cell_name_demultiplexed_with_bcl_convert", scope="session")
+def flow_cell_name_demultiplexed_with_bcl_convert() -> str:
+    return "HY7FFDRX2"
+
+
+@pytest.fixture(name="flow_cell_directory_name_demultiplexed_with_bcl_convert", scope="session")
+def fixture_flow_cell_directory_name_demultiplexed_with_bcl_convert(
+    flow_cell_name_demultiplexed_with_bcl_convert: str,
+):
+    return f"230504_A00689_0804_B{flow_cell_name_demultiplexed_with_bcl_convert}"
+
+
+@pytest.fixture(
+    name="flow_cell_directory_name_demultiplexed_with_bcl_convert_flat", scope="session"
+)
+def fixture_flow_cell_directory_name_demultiplexed_with_bcl_convert_flat(
+    flow_cell_name_demultiplexed_with_bcl_convert: str,
+):
+    """Return the name of a flow cell directory that has been demultiplexed with Bcl Convert using a flat output directory structure."""
+    return f"230505_A00689_0804_B{flow_cell_name_demultiplexed_with_bcl_convert}"
+
+
 @pytest.fixture(name="demultiplexing_init_files")
 def tmp_demultiplexing_init_files(
     bcl2fastq_flow_cell_id: str, populated_delete_demultiplex_api: DeleteDemuxAPI

tests/meta/demultiplex/test_demux_post_processing.py

@@ -1,6 +1,9 @@
 import logging
+import os
 from pathlib import Path
-from typing import Generator, List
+from typing import Dict, Generator, List
+
+import pytest
 
 
 from cg.constants.demultiplexing import BclConverter, DemultiplexingDirsAndFiles
@@ -15,6 +18,8 @@
 from cg.models.demultiplex.flow_cell import FlowCellDirectoryData
 from cg.store import Store
 
+from tests.meta.demultiplex.conftest import FlowCellInfo
+
 
 def test_set_dry_run(
     demultiplex_context: CGConfig,
@@ -485,131 +490,79 @@ def test_finish_all_flowcells(
     assert f"Check demultiplexed flow cell {bcl2fastq_flow_cell.full_name}" in caplog.text
 
 
-def test_post_processing_of_flow_cell_demultiplexed_with_bclconvert(
+@pytest.mark.parametrize(
+    "demux_type",
+    ["BCL2FASTQ_TREE", "BCLCONVERT_TREE", "BCLCONVERT_FLAT"],
+)
+def test_post_processing_of_flow_cell(
+    demux_type: str,
     demultiplex_context: CGConfig,
-    flow_cell_directory_name_demultiplexed_with_bcl_convert: str,
-    flow_cell_name_demultiplexed_with_bcl_convert: str,
+    flow_cell_info_map: Dict[str, FlowCellInfo],
     demultiplexed_flow_cells_tmp_directory: Path,
-    bcl_convert_demultiplexed_flow_cell_sample_internal_ids: List[str],
-    novaseq_6000_dir: Path,
 ):
+    """Test adding a demultiplexed flow cell to the databases with. Runs on each type of
+    demultiplexing software and setting used."""
+
+    # GIVEN a demultiplexed flow cell
+    flow_cell_demultplexing_directory: str = flow_cell_info_map.get(demux_type).directory
+    flow_cell_name: str = flow_cell_info_map.get(demux_type).name
+    sample_internal_ids: List[str] = flow_cell_info_map.get(demux_type).sample_internal_ids
+
     # GIVEN a DemuxPostProcessing API
     demux_post_processing_api = DemuxPostProcessingAPI(demultiplex_context)
 
     # GIVEN a directory with a flow cell demultiplexed with BCL Convert
     demux_post_processing_api.demux_api.out_dir = demultiplexed_flow_cells_tmp_directory
-    demux_post_processing_api.demux_api.run_dir = novaseq_6000_dir
 
-    # WHEN post processing the demultiplexed flow cell
-    demux_post_processing_api.finish_flow_cell_temp(
-        flow_cell_directory_name_demultiplexed_with_bcl_convert
-    )
-
-    # THEN a flow cell was created in statusdb
-    assert demux_post_processing_api.status_db.get_flow_cell_by_name(
-        flow_cell_name_demultiplexed_with_bcl_convert
-    )
-
-    # THEN sequencing metrics were created for the flow cell
-    assert demux_post_processing_api.status_db.get_sample_lane_sequencing_metrics_by_flow_cell_name(
-        flow_cell_name=flow_cell_name_demultiplexed_with_bcl_convert
+    # GIVEN that a sample sheet exists in the flow cell run directory
+    path = Path(
+        demux_post_processing_api.demux_api.run_dir,
+        flow_cell_demultplexing_directory,
+        DemultiplexingDirsAndFiles.SAMPLE_SHEET_FILE_NAME,
     )
-    # THEN the read count was calculated for all samples in the flow cell directory
-    for sample_id in bcl_convert_demultiplexed_flow_cell_sample_internal_ids:
-        sample = demux_post_processing_api.status_db.get_sample_by_internal_id(sample_id)
-        assert sample is not None
-        assert sample.calculated_read_count
-
-    # THEN a bundle was added to Housekeeper for the flow cell
-    assert demux_post_processing_api.hk_api.bundle(flow_cell_name_demultiplexed_with_bcl_convert)
-
-    # THEN a bundle was added to Housekeeper for each sample
-    for sample_id in bcl_convert_demultiplexed_flow_cell_sample_internal_ids:
-        assert demux_post_processing_api.hk_api.bundle(sample_id)
-
-    # THEN a sample sheet was added to Housekeeper
-    assert demux_post_processing_api.hk_api.get_files(
-        tags=[SequencingFileTag.SAMPLE_SHEET],
-        bundle=flow_cell_name_demultiplexed_with_bcl_convert,
-    ).all()
-
-    # THEN sample fastq files were added to Housekeeper tagged with FASTQ and the flow cell name
-    for sample_id in bcl_convert_demultiplexed_flow_cell_sample_internal_ids:
-        assert demux_post_processing_api.hk_api.get_files(
-            tags=[SequencingFileTag.FASTQ, flow_cell_name_demultiplexed_with_bcl_convert],
-            bundle=sample_id,
-        ).all()
-
-    # THEN a delivery file was created in the flow cell directory
-    delivery_path = Path(
-        demux_post_processing_api.demux_api.out_dir,
-        flow_cell_directory_name_demultiplexed_with_bcl_convert,
-        DemultiplexingDirsAndFiles.DELIVERY,
-    )
-
-    assert delivery_path.exists()
-
-
-def test_post_processing_of_flow_cell_demultiplexed_with_bcl2fastq(
-    demultiplex_context: CGConfig,
-    flow_cell_directory_name_demultiplexed_with_bcl2fastq: str,
-    flow_cell_name_demultiplexed_with_bcl2fastq: str,
-    demultiplexed_flow_cells_tmp_directory: Path,
-    hiseq_dir: Path,
-    bcl2fastq_demultiplexed_flow_cell_sample_internal_ids: List[str],
-):
-    # GIVEN a DemuxPostProcessing API
-    demux_post_processing_api = DemuxPostProcessingAPI(demultiplex_context)
-
-    # GIVEN a directory with a flow cell demultiplexed with bcl2fastq
-    demux_post_processing_api.demux_api.out_dir = demultiplexed_flow_cells_tmp_directory
-    demux_post_processing_api.demux_api.run_dir = hiseq_dir
+    os.makedirs(path.parent, exist_ok=True)
+    path.touch()
 
     # WHEN post processing the demultiplexed flow cell
-    demux_post_processing_api.finish_flow_cell_temp(
-        flow_cell_directory_name_demultiplexed_with_bcl2fastq
-    )
+    demux_post_processing_api.finish_flow_cell_temp(flow_cell_demultplexing_directory)
 
     # THEN a flow cell was created in statusdb
-    assert demux_post_processing_api.status_db.get_flow_cell_by_name(
-        flow_cell_name_demultiplexed_with_bcl2fastq
-    )
+    assert demux_post_processing_api.status_db.get_flow_cell_by_name(flow_cell_name)
 
     # THEN sequencing metrics were created for the flow cell
     assert demux_post_processing_api.status_db.get_sample_lane_sequencing_metrics_by_flow_cell_name(
-        flow_cell_name=flow_cell_name_demultiplexed_with_bcl2fastq
+        flow_cell_name=flow_cell_name
     )
-
     # THEN the read count was calculated for all samples in the flow cell directory
-    for sample_internal_id in bcl2fastq_demultiplexed_flow_cell_sample_internal_ids:
+    for sample_internal_id in sample_internal_ids:
         sample = demux_post_processing_api.status_db.get_sample_by_internal_id(sample_internal_id)
         assert sample is not None
         assert sample.calculated_read_count
 
     # THEN a bundle was added to Housekeeper for the flow cell
-    assert demux_post_processing_api.hk_api.bundle(flow_cell_name_demultiplexed_with_bcl2fastq)
+    assert demux_post_processing_api.hk_api.bundle(flow_cell_name)
 
     # THEN a bundle was added to Housekeeper for each sample
-    for sample_internal_id in bcl2fastq_demultiplexed_flow_cell_sample_internal_ids:
+    for sample_internal_id in sample_internal_ids:
         assert demux_post_processing_api.hk_api.bundle(sample_internal_id)
 
     # THEN a sample sheet was added to Housekeeper
     assert demux_post_processing_api.hk_api.get_files(
         tags=[SequencingFileTag.SAMPLE_SHEET],
-        bundle=flow_cell_name_demultiplexed_with_bcl2fastq,
+        bundle=flow_cell_name,
     ).all()
 
     # THEN sample fastq files were added to Housekeeper tagged with FASTQ and the flow cell name
-    for sample_internal_id in bcl2fastq_demultiplexed_flow_cell_sample_internal_ids:
+    for sample_internal_id in sample_internal_ids:
         assert demux_post_processing_api.hk_api.get_files(
-            tags=[SequencingFileTag.FASTQ, flow_cell_name_demultiplexed_with_bcl2fastq],
+            tags=[SequencingFileTag.FASTQ, flow_cell_name],
             bundle=sample_internal_id,
         ).all()
 
     # THEN a delivery file was created in the flow cell directory
     delivery_path = Path(
         demux_post_processing_api.demux_api.out_dir,
-        flow_cell_directory_name_demultiplexed_with_bcl2fastq,
+        flow_cell_demultplexing_directory,
         DemultiplexingDirsAndFiles.DELIVERY,
     )