add study plugin override (#1885)

CHANGELOG.rst

@@ -23,6 +23,7 @@ Added
     - ``notify_email_irods_request`` user app setting (#1939)
     - Assay app unit tests (#1980)
     - Missing assay plugin ``__init__.py`` files (#2014)
+    - Study plugin override via ISA-Tab comments (#1885)
 
 Changed
 -------

docs_manual/source/metadata_advanced.rst

@@ -39,6 +39,24 @@ SODAR currently supports the following study configurations:
 If the configuration is not specified or is not known to SODAR, the shortcut
 column will not be visible.
 
+It is possible to override the plugin to be used for a study. This allows for
+e.g. having different types of studies within a single investigation. The
+overriding can be done by adding a ``SODAR Study Plugin`` comment within the
+``STUDY`` section of the ISA-Tab investigation file. As the value, the full
+internal study plugin name (e.g. ``samplesheets_study_cancer``) should be used.
+If both the study plugin comment and investigation configuration comment are
+present, the former will override the latter.
+
+Example:
+
+.. code-block::
+
+    STUDY
+    Study Identifier	s_small2
+    Study Title	Small Germline Study
+    Study Description
+    Comment[SODAR Study Plugin] samplesheets_study_cancer
+
 Study plugins will search for the "latest" BAM and VCF files for shortcuts and
 IGV session generation. This is determined by file name: in case of multiple
 files, the last file sorted by file name is returned. Hence to ensure the most

docs_manual/source/sodar_release_notes.rst

@@ -16,6 +16,7 @@ Release for SODAR Core v1.0 upgrade, iRODS v4.3 upgrade and feature updates.
 - Add opt-out settings for iRODS data request and zone status update emails
 - Add REST API list view pagination
 - Add Python v3.11 support
+- Add study plugin override via ISA-Tab comments
 - Add session control in Django settings and environment variables
 - Update minimum supported iRODS version to v4.3.3
 - Update REST API versioning

samplesheets/models.py

@@ -74,7 +74,8 @@
 IRODS_REQUEST_STATUS_FAILED = 'FAILED'
 IRODS_REQUEST_STATUS_REJECTED = 'REJECTED'
 
-# ISA-Tab SODAR metadata comment key for assay plugin override
+# ISA-Tab SODAR metadata comment keys for study and assay plugin overrides
+ISA_META_STUDY_PLUGIN = 'SODAR Study Plugin'
 ISA_META_ASSAY_PLUGIN = 'SODAR Assay Plugin'
 
 
@@ -403,9 +404,19 @@ def get_plugin(self):
         """Return active study app plugin or None if not found"""
         from samplesheets.plugins import SampleSheetStudyPluginPoint
 
+        inv_config = self.investigation.get_configuration()
+        study_override = self.comments.get(ISA_META_STUDY_PLUGIN)
+        if study_override:
+            try:
+                return SampleSheetStudyPluginPoint.get_plugin(
+                    name=study_override
+                )
+            except Exception:
+                return None
         for plugin in SampleSheetStudyPluginPoint.get_plugins():
-            if plugin.config_name == self.investigation.get_configuration():
+            if plugin.config_name == inv_config:
                 return plugin
+        return None
 
     def get_url(self):
         """Return the URL for this study"""

samplesheets/studyapps/cancer/plugins.py

@@ -326,16 +326,19 @@ def update_cache(self, name=None, project=None, user=None):
                 )
             except Investigation.DoesNotExist:
                 continue
-            # Only apply for investigations with the correct configuration
-            if investigation.get_configuration() != self.config_name:
-                continue
             # If a name is given, only update that specific CacheItem
             if name:
                 study_uuid = name.split('/')[-1]
                 studies = Study.objects.filter(sodar_uuid=study_uuid)
             else:
                 studies = Study.objects.filter(investigation=investigation)
             for study in studies:
+                # Only apply for studies using this plugin
+                if (
+                    not study.get_plugin()
+                    or study.get_plugin().__class__ != self.__class__
+                ):
+                    continue
                 if self._has_only_ms_assays(study):
                     continue
                 self._update_study_cache(study, user, cache_backend)

samplesheets/studyapps/cancer/tests/test_plugins_taskflow.py

@@ -13,13 +13,14 @@
 # Taskflowbackend dependency
 from taskflowbackend.tests.base import TaskflowViewTestBase
 
+from samplesheets.models import ISA_META_STUDY_PLUGIN
+from samplesheets.plugins import SampleSheetStudyPluginPoint
 from samplesheets.rendering import SampleSheetTableBuilder
+from samplesheets.studyapps.cancer.utils import get_library_file_path
+from samplesheets.studyapps.utils import get_igv_session_url
 from samplesheets.tests.test_io import SampleSheetIOMixin, SHEET_DIR
 from samplesheets.tests.test_models import SampleSheetModelMixin
 from samplesheets.tests.test_views_taskflow import SampleSheetTaskflowMixin
-from samplesheets.plugins import SampleSheetStudyPluginPoint
-from samplesheets.studyapps.cancer.utils import get_library_file_path
-from samplesheets.studyapps.utils import get_igv_session_url
 
 
 app_settings = AppSettingAPI()
@@ -560,6 +561,18 @@ def test_update_cache(self):
             self.assertIsNone(ci['bam'][c])
             self.assertIsNone(ci['vcf'][c])
 
+    def test_update_cache_no_config(self):
+        """Test update_cache() without config"""
+        # Clear investigation configuration comments
+        self.investigation.comments = {}
+        self.investigation.save()
+        self.plugin.update_cache(self.cache_name, self.project)
+        self.assertIsNone(
+            self.cache_backend.get_cache_item(
+                APP_NAME, self.cache_name, self.project
+            )
+        )
+
     def test_update_cache_files(self):
         """Test update_cache() with files in iRODS"""
         self.irods.collections.create(self.source_path)
@@ -581,6 +594,34 @@ def test_update_cache_files(self):
             self.assertEqual(ci['bam'][CASE_IDS[i]], None)
             self.assertEqual(ci['vcf'][CASE_IDS[i]], None)
 
+    def test_update_cache_files_override(self):
+        """Test update_cache() with files in iRODS and study override"""
+        # Clear investigation configuration comments and set study override
+        self.investigation.comments = {}
+        self.investigation.save()
+        self.study.comments = {
+            ISA_META_STUDY_PLUGIN: 'samplesheets_study_cancer'
+        }
+        self.study.save()
+        self.irods.collections.create(self.source_path)
+        bam_path = os.path.join(
+            self.source_path, '{}_test.bam'.format(SAMPLE_ID_NORMAL)
+        )
+        vcf_path = os.path.join(
+            self.source_path, '{}_test.vcf.gz'.format(SAMPLE_ID_NORMAL)
+        )
+        self.irods.data_objects.create(bam_path)
+        self.irods.data_objects.create(vcf_path)
+        self.plugin.update_cache(self.cache_name, self.project)
+        ci = self.cache_backend.get_cache_item(
+            APP_NAME, self.cache_name, self.project
+        ).data
+        self.assertEqual(ci['bam'][CASE_IDS[0]], bam_path)
+        self.assertEqual(ci['vcf'][CASE_IDS[0]], vcf_path)
+        for i in range(1, len(CASE_IDS) - 1):
+            self.assertEqual(ci['bam'][CASE_IDS[i]], None)
+            self.assertEqual(ci['vcf'][CASE_IDS[i]], None)
+
     def test_update_cache_cram(self):
         """Test update_cache() with CRAM file in iRODS"""
         self.irods.collections.create(self.source_path)

samplesheets/studyapps/germline/plugins.py

@@ -424,8 +424,6 @@ def update_cache(self, name=None, project=None, user=None):
             if not investigation:
                 continue
             # Only apply for investigations with the correct configuration
-            if investigation.get_configuration() != self.config_name:
-                continue
             logger.debug(
                 'Updating cache for project {}..'.format(
                     project.get_log_title()
@@ -438,6 +436,12 @@ def update_cache(self, name=None, project=None, user=None):
             else:
                 studies = Study.objects.filter(investigation=investigation)
             for study in studies:
+                # Only apply for studies using this plugin
+                if (
+                    not study.get_plugin()
+                    or study.get_plugin().__class__ != self.__class__
+                ):
+                    continue
                 logger.debug(
                     'Updating cache for study "{}" ({})..'.format(
                         study.get_display_name(), study.sodar_uuid

samplesheets/studyapps/germline/tests/test_plugins_taskflow.py

@@ -13,14 +13,14 @@
 # Taskflowbackend dependency
 from taskflowbackend.tests.base import TaskflowViewTestBase
 
-from samplesheets.models import GenericMaterial
+from samplesheets.models import GenericMaterial, ISA_META_STUDY_PLUGIN
+from samplesheets.plugins import SampleSheetStudyPluginPoint
 from samplesheets.rendering import SampleSheetTableBuilder
+from samplesheets.studyapps.germline.utils import get_pedigree_file_path
+from samplesheets.studyapps.utils import get_igv_session_url
 from samplesheets.tests.test_io import SampleSheetIOMixin, SHEET_DIR
 from samplesheets.tests.test_models import SampleSheetModelMixin
 from samplesheets.tests.test_views_taskflow import SampleSheetTaskflowMixin
-from samplesheets.plugins import SampleSheetStudyPluginPoint
-from samplesheets.studyapps.germline.utils import get_pedigree_file_path
-from samplesheets.studyapps.utils import get_igv_session_url
 
 
 app_settings = AppSettingAPI()
@@ -594,6 +594,18 @@ def test_update_cache(self):
         self.assertIsNone(ci['vcf'][FAMILY_ID])
         self.assertIsNone(ci['vcf'][FAMILY_ID2])
 
+    def test_update_cache_no_config(self):
+        """Test update_cache() without config"""
+        # Clear investigation configuration comments
+        self.investigation.comments = {}
+        self.investigation.save()
+        self.plugin.update_cache(self.cache_name, self.project)
+        self.assertIsNone(
+            self.cache_backend.get_cache_item(
+                APP_NAME, self.cache_name, self.project
+            )
+        )
+
     def test_update_cache_files(self):
         """Test update_cache() with files in iRODS"""
         self.irods.collections.create(self.source_path)
@@ -613,6 +625,32 @@ def test_update_cache_files(self):
         self.assertEqual(ci['vcf'][FAMILY_ID], vcf_path)
         self.assertIsNone(ci['vcf'][FAMILY_ID2])
 
+    def test_update_cache_files_override(self):
+        """Test update_cache() with files in iRODS and study override"""
+        # Clear investigation configuration comments and set study override
+        self.investigation.comments = {}
+        self.investigation.save()
+        self.study.comments = {
+            ISA_META_STUDY_PLUGIN: 'samplesheets_study_germline'
+        }
+        self.study.save()
+        self.irods.collections.create(self.source_path)
+        bam_path = os.path.join(
+            self.source_path, '{}_test.bam'.format(SAMPLE_ID)
+        )
+        vcf_path = os.path.join(
+            self.source_path, '{}_test.vcf.gz'.format(FAMILY_ID)
+        )
+        self.irods.data_objects.create(bam_path)
+        self.irods.data_objects.create(vcf_path)
+        self.plugin.update_cache(self.cache_name, self.project)
+        ci = self.cache_backend.get_cache_item(
+            APP_NAME, self.cache_name, self.project
+        ).data
+        self.assertEqual(ci['bam'][self.source.name], bam_path)
+        self.assertEqual(ci['vcf'][FAMILY_ID], vcf_path)
+        self.assertIsNone(ci['vcf'][FAMILY_ID2])
+
     def test_update_cache_cram(self):
         """Test update_cache() with CRAM file in iRODS"""
         self.irods.collections.create(self.source_path)

samplesheets/tests/test_models.py

@@ -39,10 +39,12 @@
     IrodsDataRequest,
     NOT_AVAILABLE_STR,
     CONFIG_LABEL_CREATE,
+    ISA_META_STUDY_PLUGIN,
     ISA_META_ASSAY_PLUGIN,
     IRODS_REQUEST_ACTION_DELETE,
     IRODS_REQUEST_STATUS_ACTIVE,
 )
+from samplesheets.plugins import SampleSheetStudyPluginPoint
 from samplesheets.utils import get_alt_names
 
 
@@ -613,6 +615,52 @@ def test_get_url(self):
         ) + '#/study/{}'.format(self.study.sodar_uuid)
         self.assertEqual(self.study.get_url(), expected)
 
+    def test_get_plugin_unset(self):
+        """Test get_plugin() with no config set"""
+        self.assertIsNone(self.study.get_plugin())
+
+    def test_get_plugin_investigation(self):
+        """Test get_plugin() with config set in investigation"""
+        self.investigation.comments = {CONFIG_LABEL_CREATE: 'bih_germline'}
+        self.assertIsInstance(
+            self.study.get_plugin(),
+            SampleSheetStudyPluginPoint.get_plugin(
+                name='samplesheets_study_germline'
+            ).__class__,
+        )
+
+    def test_get_plugin_override(self):
+        """Test get_plugin() with config set in study override"""
+        self.study.comments = {
+            ISA_META_STUDY_PLUGIN: 'samplesheets_study_germline'
+        }
+        self.assertIsInstance(
+            self.study.get_plugin(),
+            SampleSheetStudyPluginPoint.get_plugin(
+                name='samplesheets_study_germline'
+            ).__class__,
+        )
+
+    def test_get_plugin_override_invalid(self):
+        """Test get_plugin() with invalid config name set in study override"""
+        self.study.comments = {
+            ISA_META_STUDY_PLUGIN: 'samplesheets_study_NONEXISTENT_NAME'
+        }
+        self.assertIsNone(self.study.get_plugin())
+
+    def test_get_plugin_both(self):
+        """Test get_plugin() with investigation and study configs"""
+        self.investigation.comments = {CONFIG_LABEL_CREATE: 'bih_germline'}
+        self.study.comments = {
+            ISA_META_STUDY_PLUGIN: 'samplesheets_study_cancer'
+        }
+        self.assertIsInstance(
+            self.study.get_plugin(),
+            SampleSheetStudyPluginPoint.get_plugin(
+                name='samplesheets_study_cancer'
+            ).__class__,
+        )
+
 
 class TestProtocol(SamplesheetsModelTestBase):
     """Tests for the Protocol model"""