Remove entity descriptor fields

docs/CHANGELOG.rst

@@ -16,6 +16,7 @@ Added
 
 Changed
 -------
+- **BACKWARD INCOMPATIBLE:** Require Resolwe 39.x
 - Rewrite ``expression-aggregator`` process to Python and make it
   compatible with the new annotation model
 

resolwe_bio/extensions.py

@@ -1,68 +1,10 @@
 """Resolwe-bio extensions of Resolwe core."""
 
-from django.db.models import Q
 from django_filters import rest_framework as filters
 
 from resolwe.composer import composer
 
 
-class ExtendedCollectionFilter:
-    """Collection filter extension."""
-
-    descriptor__general__species = filters.CharFilter(
-        field_name="entity__descriptor__general__species",
-        lookup_expr="icontains",
-        distinct=True,
-    )
-    tissue_type = filters.CharFilter(method="_filter_by_tissue")
-    treatment = filters.CharFilter(method="_filter_by_treatment")
-    outcome_is_defined = filters.CharFilter(method="_filter_by_outcome")
-
-    def _filter_by_multiple_fields(self, queryset, fields, value):
-        """Filter by any of the given fields.
-
-        Return exactly those objects, who icontain `value` in at least one of
-        `fields`.
-        """
-
-        query = Q()
-        for field in fields:
-            query |= Q(**{field + "__icontains": value})
-
-        return queryset.filter(query).distinct()
-
-    def _filter_by_tissue(self, queryset, name, value):
-        locations = [
-            "entity__descriptor__general__organ",
-            "entity__descriptor__general__biosample_source",
-            "entity__descriptor__disease_information__organ_part",
-            "entity__descriptor__disease_information__biopsy_site",
-            "entity__descriptor__pathological_information__organ_part",
-            "entity__descriptor__pathological_information__biopsy_site",
-        ]
-        return self._filter_by_multiple_fields(queryset, locations, value)
-
-    def _filter_by_treatment(self, queryset, name, value):
-        locations = [
-            "entity__descriptor__general__biosample_treatment",
-            "entity__descriptor__treatment_type__drug",
-            "entity__descriptor__immuno_oncology_treatment_type__io_drug",
-        ]
-        return self._filter_by_multiple_fields(queryset, locations, value)
-
-    def _filter_by_outcome(self, queryset, name, value):
-        locations = [
-            "entity__descriptor__response_and_survival_analysis__clinical_benefit",
-            "entity__descriptor__response_and_survival_analysis__confirmed_bor",
-            "entity__descriptor__response_and_survival_analysis__unconfirmed_bor",
-            "entity__descriptor__response_and_survival_analysis__pfs",
-            "entity__descriptor__response_and_survival_analysis__os",
-            "entity__descriptor__response_and_survival_analysis__dfs",
-            "entity__descriptor__response_and_survival_analysis__ttp",
-        ]
-        return self._filter_by_multiple_fields(queryset, locations, value)
-
-
 class ExtendedDataFilter:
     """Data filter extensions."""
 
@@ -81,18 +23,4 @@ def filter_output_icontains(self, queryset, name, value):
     species = filters.CharFilter(method="filter_output_icontains")
 
 
-class ExtendedEntityFilter:
-    """Data filter extensions."""
-
-    def filter_species(self, queryset, name, value):
-        """Filter queryset by genome build."""
-        return queryset.filter(descriptor__general__species__icontains=value)
-
-    species = filters.CharFilter(method="filter_species")
-
-
-composer.add_extension(
-    "resolwe.flow.filters.CollectionFilter", ExtendedCollectionFilter
-)
 composer.add_extension("resolwe.flow.filters.DataFilter", ExtendedDataFilter)
-composer.add_extension("resolwe.flow.filters.EntityFilter", ExtendedEntityFilter)

resolwe_bio/migrations/0001_squashed_0015_sample_indices.py

@@ -25,6 +25,9 @@ def load_relation_types(apps, schema_editor):
 
 class Migration(migrations.Migration):
     initial = True
+    dependencies = [
+        ("flow", "0021_annotationvalue_modified"),
+    ]
 
     operations = [
         migrations.RunPython(load_relation_types),

resolwe_bio/migrations/0005_remove_descriptor_indices.py

@@ -0,0 +1,31 @@
+# Generated by Django 3.2.12 on 2022-03-06 22:34
+import os
+
+from django.db import connection, migrations
+
+
+def load_sql(apps, schema_editor):
+    file_names = [
+        "remove_resolwe_indexes.sql",
+    ]
+    with connection.cursor() as c:
+        for file_name in file_names:
+            file_path = os.path.join(os.path.dirname(__file__), file_name)
+            with open(file_path) as fh:
+                sql_statement = fh.read()
+            c.execute(sql_statement)
+
+
+class Migration(migrations.Migration):
+    dependencies = [
+        ("resolwe_bio", "0004_merge_fastq_type"),
+        ("flow", "0021_annotationvalue_modified"),
+    ]
+
+    run_before = [
+        ("flow", "0023_remove_entity_descriptor_and_more"),
+    ]
+
+    operations = [
+        migrations.RunPython(load_sql),
+    ]

resolwe_bio/migrations/remove_resolwe_indexes.sql

@@ -0,0 +1,30 @@
+-- Delete indexes related to the entity descriptor field that has been removed.
+DROP INDEX IF EXISTS idx_entity_descriptor_sample_label_trgm,
+idx_entity_descriptor_subject_id_trgm,
+idx_entity_descriptor_batch,
+idx_entity_descriptor_group,
+idx_entity_descriptor_disease_type_trgm,
+idx_entity_descriptor_disease_status,
+idx_entity_descriptor_io_drug,
+idx_entity_descriptor_io_treatment,
+idx_entity_descriptor_confirmed_bor,
+idx_entity_descriptor_pfs_event,
+idx_entity_descriptor_description_trgm,
+idx_entity_descriptor_biosample_source_trgm,
+idx_entity_descriptor_biosample_treatment_trgm,
+idx_entity_descriptor__general__organ,
+idx_entity_descriptor__general__biosample_source,
+idx_entity_descriptor__disease_information__organ_part,
+idx_entity_descriptor__disease_information__biopsy_site,
+idx_entity_descriptor__pathological_information__organ_part,
+idx_entity_descriptor__pathological_information__biopsy_site,
+idx_entity_descriptor__general__biosample_treatment,
+idx_entity_descriptor__treatment_type__drug,
+idx_entity_descriptor__immuno_oncology_treatment_type__io_drug,
+idx_entity_descriptor__response_and_survival_analysis__clinical_benefit,
+idx_entity_descriptor__response_and_survival_analysis__confirmed_bor,
+idx_entity_descriptor__response_and_survival_analysis__unconfirmed_bor,
+idx_entity_descriptor__response_and_survival_analysis__pfs,
+idx_entity_descriptor__response_and_survival_analysis__os,
+idx_entity_descriptor__response_and_survival_analysis__dfs,
+idx_entity_descriptor__response_and_survival_analysis__ttp;

resolwe_bio/migrations/resolwe_indexes.sql

@@ -53,36 +53,3 @@ CREATE OR REPLACE FUNCTION data_biut()
         RETURN NEW;
     END;
     $$;
-
--- Add resolwe-bio specific fields to Entity search.
-CREATE OR REPLACE FUNCTION generate_resolwe_bio_entity_search(entity flow_entity)
-    RETURNS tsvector
-    LANGUAGE plpgsql
-    AS $$
-    DECLARE
-        search tsvector;
-    BEGIN
-        SELECT
-            -- Species.
-            setweight(to_tsvector('simple', COALESCE(entity.descriptor->'general'->>'species', '')), 'A')
-        INTO search;
-
-        RETURN search;
-    END;
-    $$;
-
-CREATE OR REPLACE FUNCTION entity_biut()
-    RETURNS TRIGGER
-    LANGUAGE plpgsql
-    AS $$
-    BEGIN
-        SELECT generate_resolwe_entity_search(NEW) || generate_resolwe_bio_entity_search(NEW)
-        INTO NEW.search;
-
-        IF NEW.search IS NULL THEN
-            RAISE WARNING 'Search index for entity (id: %) is NULL.', NEW.id;
-        END IF;
-
-        RETURN NEW;
-    END;
-    $$;

resolwe_bio/tests/processes/test_chipseq.py

@@ -625,7 +625,6 @@ def set_sample_name(data, sample_name):
                 contributor=self.contributor,
                 type="data:chipseq:callpeak:macs2:",
                 entity_type="sample",
-                entity_descriptor_schema="sample",
                 input_schema=[
                     {
                         "name": "src",
@@ -679,7 +678,6 @@ def set_sample_name(data, sample_name):
                 contributor=self.contributor,
                 type="data:chipseq:callpeak:macs14:",
                 entity_type="sample",
-                entity_descriptor_schema="sample",
                 input_schema=[
                     {
                         "name": "src",

resolwe_bio/tests/processes/test_diff_expression.py

@@ -203,7 +203,6 @@ def test_deseq2_nanostring(self):
                 contributor=self.contributor,
                 type="data:expression:nanostring:",
                 entity_type="sample",
-                entity_descriptor_schema="sample",
                 data_name="{{ src.file }}",
                 input_schema=[
                     {
@@ -311,7 +310,6 @@ def test_de_microarray(self):
                 contributor=self.contributor,
                 type="data:expression:microarray:",
                 entity_type="sample",
-                entity_descriptor_schema="sample",
                 data_name="{{ name }}",
                 input_schema=[
                     {

resolwe_bio/tests/processes/test_support_processors.py

@@ -185,7 +185,6 @@ def test_archive_samples_exp(self):
                 requirements={"expression-engine": "jinja"},
                 data_name="Upload expression into exp output field",
                 entity_type="sample",
-                entity_descriptor_schema="sample",
                 input_schema=[
                     {
                         "name": "exp",
@@ -545,7 +544,6 @@ def set_sample_name(data, sample_name):
                     },
                 },
                 entity_type="sample",
-                entity_descriptor_schema="sample",
                 contributor=self.contributor,
                 type="data:wgbs:bsrate:",
                 input_schema=[
@@ -617,7 +615,6 @@ def set_sample_name(data, sample_name):
                     },
                 },
                 entity_type="sample",
-                entity_descriptor_schema="sample",
                 contributor=self.contributor,
                 type="data:alignment:bam:walt:",
                 input_schema=[
@@ -685,7 +682,6 @@ def set_sample_name(data, sample_name):
                     },
                 },
                 entity_type="sample",
-                entity_descriptor_schema="sample",
                 contributor=self.contributor,
                 type="data:chipqc:",
                 input_schema=[
@@ -747,7 +743,6 @@ def set_sample_name(data, sample_name):
                     },
                 },
                 entity_type="sample",
-                entity_descriptor_schema="sample",
                 contributor=self.contributor,
                 type="data:chipseq:callpeak:macs2:",
                 input_schema=[
@@ -834,7 +829,6 @@ def set_sample_name(data, sample_name):
                     },
                 },
                 entity_type="sample",
-                entity_descriptor_schema="sample",
                 contributor=self.contributor,
                 type="data:nanostring:rcc:",
                 input_schema=[

resolwe_bio/tests/processes/test_wgs.py

@@ -183,7 +183,6 @@ def test_gatk_genomicsdb(self):
                 contributor=self.contributor,
                 type="data:variants:gvcf:",
                 entity_type="sample",
-                entity_descriptor_schema="sample",
                 data_name="{{ gvcf.file }}",
                 input_schema=[
                     {
@@ -310,7 +309,6 @@ def test_gatk_genotypegvcfs(self):
                 contributor=self.contributor,
                 type="data:variants:gvcf:",
                 entity_type="sample",
-                entity_descriptor_schema="sample",
                 data_name="{{ gvcf.file }}",
                 input_schema=[
                     {