Merge pull request #365 from EBI-Metagenomics/add-new-analysis-files

Add new analysis files

emgapi/migrations/0018_genome_annotations_v2_4_downloads.py

@@ -0,0 +1,48 @@
+# Generated by Django 3.2.12 on 2024-08-15 00:00
+
+from django.db import migrations
+
+
+def create_v2_4_genome_download_description(apps, schema_editor):
+    DownloadDescriptionLabel = apps.get_model("emgapi", "DownloadDescriptionLabel")
+
+    DownloadDescriptionLabel.objects.create(
+        description="CAZy and putative polysaccharide utilization loci annotations produced by dbCAN",
+        description_label="Genome dbCAN Annotation"
+    )
+    DownloadDescriptionLabel.objects.create(
+        description="Anti-phage defence system annotations produced by Defense Finder",
+        description_label="Genome Defense Finder Annotation"
+    )
+    DownloadDescriptionLabel.objects.create(
+        description="BGC annotations produced by GECCO",
+        description_label="Genome GECCO Annotation"
+    )
+    DownloadDescriptionLabel.objects.create(
+        description="Completeness of identified KEGG pathways",
+        description_label="KEGG Pathway Completeness"
+    )
+
+
+def remove_v2_4_genome_download_description(apps, schema_editor):
+    DownloadDescriptionLabel = apps.get_model("emgapi", "DownloadDescriptionLabel")
+    DownloadDescriptionLabel.objects.filter(description_label__in=[
+        "Genome dbCAN Annotation",
+        "Genome Defense Finder Annotation",
+        "Genome GECCO Annotation",
+        "KEGG Pathway Completeness",
+    ]).delete()
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('emgapi', '0017_auto_20240129_1401'),
+    ]
+
+    operations = [
+        migrations.RunPython(
+            code=create_v2_4_genome_download_description,
+            reverse_code=remove_v2_4_genome_download_description,
+        ),
+    ]

emgapianns/management/commands/import_genomes.py

@@ -387,6 +387,15 @@ def upload_genome_files(self, genome, directory, has_pangenome):
                                 genome.accession + '_crisprcasfinder.tsv', 'Genome analysis', 'genome', False)
         self.upload_genome_file(genome, directory, 'Genome Mobilome Annotation', 'gff',
                                 genome.accession + '_mobilome.gff', 'Genome analysis', 'genome', False)
+        # pipeline v2.4.0 files (if not empty):
+        self.upload_genome_file(genome, directory, 'Genome dbCAN Annotation', 'gff',
+                                genome.accession + '_dbcan.gff', 'Genome analysis', 'genome', False)
+        self.upload_genome_file(genome, directory, 'Genome Defense Finder Annotation', 'gff',
+                                genome.accession + '_defense_finder.gff', 'Genome analysis', 'genome', False)
+        self.upload_genome_file(genome, directory, 'Genome GECCO Annotation', 'gff',
+                                genome.accession + '_gecco.gff', 'Genome analysis', 'genome', False)
+        self.upload_genome_file(genome, directory, 'KEGG Pathway Completeness', 'tsv',
+                                genome.accession + '_kegg_pathways.tsv', 'Genome analysis', 'genome', False)
 
         if has_pangenome:
             self.upload_genome_file(genome, directory, 'Pangenome core genes list', 'tab',

emgcli/__init__.py

@@ -1 +1 @@
-__version__: str = "2.4.48"
+__version__: str = "2.4.49"

pyproject.toml

@@ -117,7 +117,7 @@ max-line-length = 119
 """
 
 [tool.bumpversion]
-current_version = "2.4.48"
+current_version = "2.4.49"
 
 [[tool.bumpversion.files]]
 filename = "emgcli/__init__.py"

tests/webuploader/test_genomes.py

@@ -65,6 +65,10 @@ def _setup(self):
             ("Additional data for CRISPRCasFinder records reported in the CRISPRCasFinder GFF", "Genome CRISPRCasFinder Additional Records"),
             ("Annotated viral sequence and mobile elements", "Genome Mobilome Annotation"),
             ("List of genes in the pan-genome with their annotation and MGYG accessions.", "Gene Presence / Absence list"),
+            ("CAZy and putative polysaccharide utilization loci annotations produced by dbCAN", "Genome dbCAN Annotation"),
+            ("Anti-phage defence system annotations produced by Defense Finder", "Genome Defense Finder Annotation"),
+            ("BGC annotations produced by GECCO", "Genome GECCO Annotation"),
+            ("Completeness of identified KEGG pathways", "KEGG Pathway Completeness"),
         )
         for d in downloads:
             emg_models.DownloadDescriptionLabel.objects.get_or_create(