updates data model, importer, api, filters for first release of viral…

… catalogues
EBI-Metagenomics · Jan 17, 2024 · 544c255 · 544c255
1 parent 56c34d0
commit 544c255
Show file tree

Hide file tree

Showing 9 changed files with 67 additions and 50 deletions.
diff --git a/holofood/api.py b/holofood/api.py
@@ -209,7 +209,6 @@ class Config:
 
 class ViralFragmentSchema(ModelSchema):
     cluster_representative: Optional["ViralFragmentSchema"]
-    host_mag: Optional[GenomeSchema]
 
     @staticmethod
     def resolve_contig_url(obj: ViralFragment):
@@ -238,8 +237,8 @@ class Config:
             "start_within_contig",
             "end_within_contig",
             "metadata",
-            "host_mag",
             "viral_type",
+            "taxonomy",
         ]
 
 
@@ -483,9 +482,7 @@ def get_viral_catalogue(request, catalogue_id: str):
     "found in the assembly contigs of HoloFood samples."
     "The Catalogue’s viral fragments are all from the same biome."
     "Viral sequences are clustered by sequence identity, at a species-level."
-    "Both cluster representatives and cluster members are included."
-    "Where a viral sequence is found in a related MAG (metagenome assembly genome,"
-    " e.g. a bacterial species), this MAG is considered a “host MAG”.",
+    "Both cluster representatives and cluster members are included.",
     tags=[VIRUSES],
 )
 def list_viral_catalogue_fragments(request, catalogue_id: str):

diff --git a/holofood/filters.py b/holofood/filters.py
@@ -145,8 +145,7 @@ class Meta:
             "id": ["icontains"],
             "contig_id": ["icontains"],
             "viral_type": ["exact"],
-            "host_mag__taxonomy": ["icontains"],
-            "host_mag__accession": ["icontains"],
+            "taxonomy": ["icontains"],
         }
 
     def cluster_representative_status(self, queryset, name, value):

diff --git a/holofood/management/commands/import_viral_catalogue.py b/holofood/management/commands/import_viral_catalogue.py
@@ -94,7 +94,6 @@ def handle(self, *args, **options):
             "sequence_end": "end_within_contig",
             "cluster_representative": "cluster_representative_id",
             "mgya": "mgnify_analysis_accession",
-            "host_mgyg": "host_mag_id",
             "viral_type": "viral_type",
             "contig": "contig_id",
             "viral_taxonomy": "taxonomy",
@@ -130,14 +129,14 @@ def handle(self, *args, **options):
                 for col_name, field_name in column_mapping.items()
                 if sequence[col_name] != ""
             }
-
-            vir_data["taxonomy"] = (
-                vir_data["taxonomy"]
-                .replace(";", " > ")
-                .strip()
-                .lstrip(" >")
-                .rstrip(" >")
-            )
+            if "taxonomy" in vir_data:
+                vir_data["taxonomy"] = (
+                    vir_data["taxonomy"]
+                    .replace(";", " > ")
+                    .strip()
+                    .lstrip(" >")
+                    .rstrip(" >")
+                )
 
             contig_id = sequence["contig"]
 

diff --git a/holofood/migrations/0034_alter_animal_options_alter_viralfragment_options_and_more.py b/holofood/migrations/0034_alter_animal_options_alter_viralfragment_options_and_more.py
@@ -0,0 +1,24 @@
+# Generated by Django 4.2 on 2024-01-17 12:57
+
+from django.db import migrations
+
+
+class Migration(migrations.Migration):
+    dependencies = [
+        ("holofood", "0033_remove_sample_ena_run_accessions_and_more"),
+    ]
+
+    operations = [
+        migrations.AlterModelOptions(
+            name="animal",
+            options={"ordering": ("accession",)},
+        ),
+        migrations.AlterModelOptions(
+            name="viralfragment",
+            options={"ordering": ("id",)},
+        ),
+        migrations.RemoveField(
+            model_name="viralfragment",
+            name="host_mag",
+        ),
+    ]
diff --git a/holofood/migrations/0035_alter_viralfragment_viral_type.py b/holofood/migrations/0035_alter_viralfragment_viral_type.py
@@ -0,0 +1,23 @@
+# Generated by Django 4.2 on 2024-01-17 15:12
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+    dependencies = [
+        ("holofood", "0034_alter_animal_options_alter_viralfragment_options_and_more"),
+    ]
+
+    operations = [
+        migrations.AlterField(
+            model_name="viralfragment",
+            name="viral_type",
+            field=models.CharField(
+                choices=[
+                    ("viral_sequence", "viral_sequence"),
+                    ("prophage", "prophage"),
+                ],
+                max_length=15,
+            ),
+        ),
+    ]
diff --git a/holofood/models.py b/holofood/models.py
@@ -491,8 +491,8 @@ class ViralFragment(models.Model):
     objects = ViralFragmentClusterManager()
 
     PROPHAGE = "prophage"
-    PHAGE = "phage"
-    VIRAL_TYPE_CHOICES = [(PHAGE, PHAGE), (PROPHAGE, PROPHAGE)]
+    VIRAL_SEQUENCE = "viral_sequence"
+    VIRAL_TYPE_CHOICES = [(VIRAL_SEQUENCE, VIRAL_SEQUENCE), (PROPHAGE, PROPHAGE)]
 
     id = models.CharField(primary_key=True, max_length=100, verbose_name="ID")
     catalogue = models.ForeignKey(
@@ -510,15 +510,7 @@ class ViralFragment(models.Model):
     start_within_contig = models.IntegerField()
     end_within_contig = models.IntegerField()
     metadata = models.JSONField(default=dict, blank=True)
-    host_mag = models.ForeignKey(
-        Genome,
-        on_delete=models.SET_NULL,
-        null=True,
-        blank=True,
-        related_name="viral_fragments",
-        verbose_name="Host MAG",
-    )
-    viral_type = models.CharField(choices=VIRAL_TYPE_CHOICES, max_length=10)
+    viral_type = models.CharField(choices=VIRAL_TYPE_CHOICES, max_length=15)
     taxonomy = models.CharField(null=True, blank=True, max_length=100)
 
     gff = models.TextField(blank=True, default="")

diff --git a/holofood/templatetags/taxonomy.py b/holofood/templatetags/taxonomy.py
@@ -5,6 +5,8 @@
 
 @register.inclusion_tag("holofood/components/atoms/taxonomy_tooltip.html")
 def taxonomy_tooltip(taxonomy: str, separator: str = ">", default: str = "—") -> dict:
-    if taxonomy is None or separator not in taxonomy:
+    if not taxonomy:
         return {"short_tax": default, "long_tax": None}
+    if separator not in taxonomy:
+        return {"short_tax": taxonomy, "long_tax": None}
     return {"short_tax": taxonomy.split(separator)[-1].strip(), "long_tax": taxonomy}
diff --git a/holofood/tests/static_fixtures/viral_catalogue/viral_cat.tsv b/holofood/tests/static_fixtures/viral_catalogue/viral_cat.tsv
@@ -1,3 +1,3 @@
-sequence_id	contig	viral_type	viral_taxonomy	host_mgyg	sequence_start	sequence_end	cluster_representative	mgya
-MGYC001-start-256-end-512	MGYC001	prophage	Caudovirales	MGYG999	256	512		MGYA001
-MGYC003-start-512-end-768	MGYC003	prophage	Caudovirales	MGYG999	512	768	MGYC001-start-256-end-512	MGYA002
+sequence_id	contig	viral_type	viral_taxonomy	sequence_start	sequence_end	cluster_representative	mgya
+MGYC001-start-256-end-512	MGYC001	prophage	Caudovirales	256	512		MGYA001
+MGYC003-start-512-end-768	MGYC003	prophage	Caudovirales	512	768	MGYC001-start-256-end-512	MGYA002
diff --git a/templates/holofood/pages/viral_catalogue_detail.html b/templates/holofood/pages/viral_catalogue_detail.html
@@ -124,8 +124,6 @@ <h2>{{ catalogue.title }} viral catalogue</h2>
                             <th class="vf-table__heading" scope="col">Parent contig</th>
                             <th class="vf-table__heading" scope="col">Type</th>
                             <th class="vf-table__heading" scope="col">Taxonomy</th>
-                            <th class="vf-table__heading" scope="col">Host MAG</th>
-                            <th class="vf-table__heading" scope="col">Host taxonomy</th>
                             <th class="vf-table__heading" scope="col">View</th>
                         </tr>
                     </thead>
@@ -141,23 +139,6 @@ <h2>{{ catalogue.title }} viral catalogue</h2>
                                     {{ viral_fragment.viral_type }}
                                 </td>
                                 <td class="vf-table__cell">{% taxonomy_tooltip taxonomy=viral_fragment.taxonomy %}</td>
-                                <td class="vf-table__cell">
-                                    {% if viral_fragment.host_mag %}
-                                        <a class="vf-button vf-button--link vf-button--sm inline"
-                                           href="{% url 'genome_catalogue' pk=catalogue.related_genome_catalogue.id %}?accession__icontains={{ viral_fragment.host_mag.accession }}">
-                                            {{ viral_fragment.host_mag.accession }}
-                                        </a>
-                                    {% else %}
-                                        —
-                                    {% endif %}
-                                </td>
-                                <td class="vf-table__cell">
-                                    {% if viral_fragment.host_mag %}
-                                        {% taxonomy_tooltip taxonomy=viral_fragment.host_mag.taxonomy %}
-                                    {% else %}
-                                        —
-                                    {% endif %}
-                                </td>
                                 <td class="vf-table__cell">
                                     {% if viral_fragment.id == selected_viral_fragment.id %}
                                         <b>Viewing</b>