Remove pbi file from post processing

cg/services/post_processing/pacbio/run_file_manager/run_file_manager.py

@@ -14,19 +14,19 @@ class PacBioRunFileManager(RunFileManager):
     def get_files_to_parse(self, run_data: PacBioRunData) -> list[Path]:
         """Get the file paths required by the PacBioMetricsParser."""
         run_path: Path = run_data.full_path
-        files_to_parse: list[Path] = [self._find_ccs_report_file(run_path)]
-        files_to_parse.extend(self._get_report_files(run_path))
+        files_to_parse: list[Path] = self._get_report_files(run_path)
+        files_to_parse.append(self._get_ccs_report_file(run_path))
         return files_to_parse
 
     def get_files_to_store(self, run_data: PacBioRunData) -> list[Path]:
         """Get the files to store for the PostProcessingHKService."""
         run_path: Path = run_data.full_path
-        files_to_store: list[Path] = self._find_hifi_files(run_path)
-        files_to_store.extend(self.get_files_to_parse(run_data))
+        files_to_store: list[Path] = self.get_files_to_parse(run_data)
+        files_to_store.append(self._get_hifi_read_file(run_path))
         return files_to_store
 
     @staticmethod
-    def _find_ccs_report_file(run_path: Path) -> Path:
+    def _get_ccs_report_file(run_path: Path) -> Path:
         """Return the path to the the CCS report file."""
         statistics_dir: Path = Path(run_path, PacBioDirsAndFiles.STATISTICS_DIR)
         files: list[Path] = get_files_matching_pattern(
@@ -52,15 +52,11 @@ def _get_report_files(run_path: Path) -> list[Path]:
         return report_files
 
     @staticmethod
-    def _find_hifi_files(run_path: Path) -> list[Path]:
-        """Return the paths to the HiFi read files."""
+    def _get_hifi_read_file(run_path: Path) -> Path:
+        """Return the path to the HiFi read file."""
         hifi_dir = Path(run_path, PacBioDirsAndFiles.HIFI_READS)
-        bam_files: list[Path] = get_files_matching_pattern(
-            directory=hifi_dir, pattern=f"*{FileExtensions.BAM}*"
-        )
-        if len(bam_files) != 2:
-            raise PostProcessingFileNotFoundError(
-                f"Expected 2 HiFi read files in {hifi_dir}, found {len(bam_files)}"
-            )
-        validate_files_exist(bam_files)
-        return bam_files
+        bam_file: Path = get_files_matching_pattern(
+            directory=hifi_dir, pattern=f"*{FileExtensions.BAM}"
+        )[0]
+        validate_files_exist([bam_file])
+        return bam_file

tests/fixture_plugins/pacbio_fixtures/path_fixtures.py

@@ -110,20 +110,15 @@ def pac_bio_report_files_to_parse(
 ) -> list[Path]:
     """Return the list of PacBio report files to parse."""
     return [
-        pac_bio_ccs_report_file,
         pac_bio_control_report_file,
         pac_bio_loading_report_file,
         pac_bio_raw_data_report_file,
         pac_bio_smrtlink_datasets_report_file,
+        pac_bio_ccs_report_file,
     ]
 
 
 @pytest.fixture
-def pac_bio_hifi_files(
-    pac_bio_hifi_reads_dir: Path, pac_bio_1_a01_cell_full_name: str
-) -> list[Path]:
-    """Return the list of PacBio HiFi read files."""
-    return [
-        Path(pac_bio_hifi_reads_dir, f"{pac_bio_1_a01_cell_full_name}.hifi_reads.bam"),
-        Path(pac_bio_hifi_reads_dir, f"{pac_bio_1_a01_cell_full_name}.hifi_reads.bam.pbi"),
-    ]
+def pac_bio_hifi_read_file(pac_bio_hifi_reads_dir: Path, pac_bio_1_a01_cell_full_name: str) -> Path:
+    """Return the PacBio HiFi read file."""
+    return Path(pac_bio_hifi_reads_dir, f"{pac_bio_1_a01_cell_full_name}.hifi_reads.bam")

tests/services/post_processing/pacbio/run_file_manager/test_pacbio_run_file_manager.py

@@ -26,7 +26,7 @@ def test_get_files_to_parse(
 def test_get_files_to_store(
     expected_pac_bio_run_data: PacBioRunData,
     pac_bio_report_files_to_parse: list[Path],
-    pac_bio_hifi_files: list[Path],
+    pac_bio_hifi_read_file: Path,
 ):
     """Test that the files to be stored are returned"""
 
@@ -39,5 +39,5 @@ def test_get_files_to_store(
     files: list[Path] = file_manager.get_files_to_store(expected_pac_bio_run_data)
 
     # THEN the correct files are returned
-    full_list: list[Path] = pac_bio_report_files_to_parse + pac_bio_hifi_files
+    full_list: list[Path] = pac_bio_report_files_to_parse + [pac_bio_hifi_read_file]
     assert set(files) == set(full_list)