Add variants application

resolwe_bio/variants/__init__.py

@@ -0,0 +1,7 @@
+""".. Ignore pydocstyle D400.
+
+===================================
+Resolwe Bioinformatics Variants App
+===================================
+
+"""

resolwe_bio/variants/apps.py

@@ -0,0 +1,16 @@
+""".. Ignore pydocstyle D400.
+
+===============================
+Variants Base App Configuration
+===============================
+
+"""
+from django.apps import AppConfig
+
+
+class VariantsConfig(AppConfig):
+    """App configuration."""
+
+    name = "resolwe_bio.variants"
+    label = "resolwe_bio_variants"
+    verbose_name = "Resolwe Bioinformatics Variants Base"

resolwe_bio/variants/listener_plugin.py

@@ -0,0 +1,75 @@
+"""Handle variants related commands."""
+
+import logging
+from typing import TYPE_CHECKING
+
+from resolwe.flow.executors.socket_utils import Message, Response
+from resolwe.flow.managers.listener.plugin import (
+    ListenerPlugin,
+    listener_plugin_manager,
+)
+
+from .models import Variant, VariantCall, VariantExperiment
+
+if TYPE_CHECKING:
+    from resolwe.flow.managers.listener.listener import Processor
+
+logger = logging.getLogger(__name__)
+
+
+class VariantCommands(ListenerPlugin):
+    """Listener handlers related to the variants application."""
+
+    plugin_manager = listener_plugin_manager
+
+    def add_variants(
+        self, data_id: int, message: Message[dict], manager: "Processor"
+    ) -> Response[int]:
+        """Handle connecting variants with the samples.
+
+        If the reported variant does not exist in the file it is created.
+        """
+        data = manager.data(data_id)
+        sample = data.entity
+        metadata, variants_data = message.message_data
+        species, genome_assembly = metadata["species"], metadata["genome_assembly"]
+
+        variant_calls = list()
+        variant_cache = dict()
+        experiment = VariantExperiment.objects.create(
+            variant_data_source=metadata["variant_data_source"],
+            contributor=data.contributor,
+        )
+
+        # Bulk create variants. The consequesce of ignore_conflicts flag is that the
+        # database does not returt the ids of the created objects. So first create all
+        # the variants and then create the variant calls.
+        for variant_data in variants_data:
+            key = {
+                "species": species,
+                "genome_assembly": genome_assembly,
+                "chromosome": variant_data["chromosome"],
+                "position": variant_data["position"],
+                "reference": variant_data["reference"],
+                "alternative": variant_data["alternative"],
+            }
+            # To reduce the hits to the database use cache for variants.
+            key_tuple = tuple(key.values())
+            if key_tuple not in variant_cache:
+                variant_cache[key_tuple] = Variant.objects.get_or_create(**key)[0]
+            variant = variant_cache[key_tuple]
+
+            variant_calls.append(
+                VariantCall(
+                    variant=variant,
+                    data=data,
+                    sample=sample,
+                    quality=variant_data["quality"],
+                    depth=variant_data["depth"],
+                    genotype=variant_data["genotype"],
+                    filter=variant_data["filter"],
+                    experiment=experiment,
+                )
+            )
+
+        VariantCall.objects.bulk_create(variant_calls)

resolwe_bio/variants/migrations/0001_initial.py

@@ -0,0 +1,192 @@
+# Generated by Django 4.2.11 on 2024-03-25 10:07
+
+from django.conf import settings
+import django.contrib.postgres.fields
+from django.db import migrations, models
+import django.db.models.deletion
+
+
+class Migration(migrations.Migration):
+
+    initial = True
+
+    dependencies = [
+        migrations.swappable_dependency(settings.AUTH_USER_MODEL),
+        ("flow", "0021_annotationvalue_modified"),
+    ]
+
+    operations = [
+        migrations.CreateModel(
+            name="Variant",
+            fields=[
+                (
+                    "id",
+                    models.AutoField(
+                        auto_created=True,
+                        primary_key=True,
+                        serialize=False,
+                        verbose_name="ID",
+                    ),
+                ),
+                ("species", models.CharField(max_length=50)),
+                ("genome_assembly", models.CharField(max_length=20)),
+                ("chromosome", models.CharField(max_length=20)),
+                ("position", models.PositiveBigIntegerField()),
+                ("reference", models.CharField(max_length=100)),
+                ("alternative", models.CharField(max_length=100)),
+            ],
+        ),
+        migrations.CreateModel(
+            name="VariantExperiment",
+            fields=[
+                (
+                    "id",
+                    models.AutoField(
+                        auto_created=True,
+                        primary_key=True,
+                        serialize=False,
+                        verbose_name="ID",
+                    ),
+                ),
+                ("variant_data_source", models.CharField(max_length=100)),
+                ("date", models.DateTimeField(auto_now_add=True, db_index=True)),
+                (
+                    "contributor",
+                    models.ForeignKey(
+                        on_delete=django.db.models.deletion.PROTECT,
+                        to=settings.AUTH_USER_MODEL,
+                    ),
+                ),
+            ],
+        ),
+        migrations.CreateModel(
+            name="VariantCall",
+            fields=[
+                (
+                    "id",
+                    models.AutoField(
+                        auto_created=True,
+                        primary_key=True,
+                        serialize=False,
+                        verbose_name="ID",
+                    ),
+                ),
+                ("quality", models.FloatField()),
+                ("depth", models.PositiveIntegerField()),
+                ("filter", models.CharField(max_length=20)),
+                ("genotype", models.CharField(blank=True, max_length=100, null=True)),
+                (
+                    "data",
+                    models.ForeignKey(
+                        blank=True,
+                        null=True,
+                        on_delete=django.db.models.deletion.CASCADE,
+                        related_name="variant_calls",
+                        to="flow.data",
+                    ),
+                ),
+                (
+                    "experiment",
+                    models.ForeignKey(
+                        blank=True,
+                        null=True,
+                        on_delete=django.db.models.deletion.CASCADE,
+                        related_name="variant_calls",
+                        to="resolwe_bio_variants.variantexperiment",
+                    ),
+                ),
+                (
+                    "sample",
+                    models.ForeignKey(
+                        blank=True,
+                        null=True,
+                        on_delete=django.db.models.deletion.CASCADE,
+                        related_name="variant_calls",
+                        to="flow.entity",
+                    ),
+                ),
+                (
+                    "variant",
+                    models.ForeignKey(
+                        blank=True,
+                        null=True,
+                        on_delete=django.db.models.deletion.CASCADE,
+                        related_name="variant_calls",
+                        to="resolwe_bio_variants.variant",
+                    ),
+                ),
+            ],
+        ),
+        migrations.CreateModel(
+            name="VariantAnnotation",
+            fields=[
+                (
+                    "id",
+                    models.AutoField(
+                        auto_created=True,
+                        primary_key=True,
+                        serialize=False,
+                        verbose_name="ID",
+                    ),
+                ),
+                ("type", models.CharField(blank=True, max_length=100, null=True)),
+                ("annotation", models.CharField(max_length=200)),
+                ("annotation_impact", models.CharField(max_length=20)),
+                ("gene", models.CharField(max_length=100)),
+                ("protein_impact", models.CharField(max_length=100)),
+                (
+                    "feature_id",
+                    django.contrib.postgres.fields.ArrayField(
+                        base_field=models.CharField(max_length=200),
+                        default=list,
+                        size=None,
+                    ),
+                ),
+                (
+                    "clinical_diagnosis",
+                    models.CharField(blank=True, max_length=200, null=True),
+                ),
+                (
+                    "clinical_significance",
+                    models.CharField(blank=True, max_length=100, null=True),
+                ),
+                ("dbsnp_id", models.CharField(blank=True, max_length=20, null=True)),
+                (
+                    "clinical_var_id",
+                    models.CharField(blank=True, max_length=20, null=True),
+                ),
+                (
+                    "data",
+                    models.ForeignKey(
+                        blank=True,
+                        null=True,
+                        on_delete=django.db.models.deletion.CASCADE,
+                        related_name="variant_annotations",
+                        to="flow.data",
+                    ),
+                ),
+                (
+                    "variant",
+                    models.OneToOneField(
+                        on_delete=django.db.models.deletion.CASCADE,
+                        related_name="annotation",
+                        to="resolwe_bio_variants.variant",
+                    ),
+                ),
+            ],
+        ),
+        migrations.AddConstraint(
+            model_name="variant",
+            constraint=models.UniqueConstraint(
+                fields=(
+                    "species",
+                    "genome_assembly",
+                    "chromosome",
+                    "position",
+                    "reference",
+                    "alternative",
+                ),
+                name="uniq_composite_key_variants",
+            ),
+        ),
+    ]