#1224 - VEP - structural variants too long to annotate

annotation/migrations/0124_alter_variantannotation_vep_skipped_reason.py

@@ -0,0 +1,18 @@
+# Generated by Django 4.2.18 on 2025-01-28 23:32
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('annotation', '0123_one_off_fix_annotation_sv_c_hgvs'),
+    ]
+
+    operations = [
+        migrations.AlterField(
+            model_name='variantannotation',
+            name='vep_skipped_reason',
+            field=models.CharField(blank=True, choices=[('c', 'Unknown Contig'), ('i', 'Incomplete'), ('u', 'Unknown'), ('l', 'Too Long')], max_length=1, null=True),
+        ),
+    ]

annotation/migrations/0125_one_off_historical_vep_too_long.py

@@ -0,0 +1,31 @@
+# Generated by Django 4.2.18 on 2025-01-29 00:48
+import logging
+
+from django.db import migrations
+from django.db.models.functions import Abs
+
+
+def _one_off_historical_vep_too_long(apps, schema_editor):
+    Variant = apps.get_model("snpdb", "Variant")
+    VariantAnnotation = apps.get_model("annotation", "VariantAnnotation")
+
+    VEP_SKIPPED_UNKNOWN = 'u'
+    VEP_SKIPPED_TOO_LONG = 'l'
+
+    qs = Variant.objects.filter(svlen__isnull=False).annotate(abs_svlen=Abs("svlen"))
+    qs_long_variants = qs.filter(abs_svlen__gte=10_000_000)
+    qs_qa_long = VariantAnnotation.objects.filter(variant__in=qs_long_variants,
+                                                  vep_skipped_reason=VEP_SKIPPED_UNKNOWN)
+    if ret := qs_qa_long.update(vep_skipped_reason=VEP_SKIPPED_TOO_LONG):
+        logging.info("Updated %d VEP skipped variants to TOO_LONG ", ret)
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('annotation', '0124_alter_variantannotation_vep_skipped_reason'),
+    ]
+
+    operations = [
+        migrations.RunPython(_one_off_historical_vep_too_long)
+    ]

annotation/models/models_enums.py

@@ -158,6 +158,7 @@ class VEPSkippedReason(models.TextChoices):
     UNKNOWN_CONTIG = 'c', "Unknown Contig"
     INCOMPLETE = 'i', "Incomplete"
     UNKNOWN = 'u', "Unknown"
+    TOO_LONG = 'l', "Too Long"
 
 
 class ClinVarReviewStatus(models.TextChoices):

annotation/vcf_files/import_vcf_annotations.py

@@ -84,6 +84,7 @@ def handle_vep_warnings(annotation_run: AnnotationRun, bulk_inserter):
 
         version = annotation_run.annotation_range_lock.version
         annotation_version = version.get_any_annotation_version()
+        # This pulls down any un-annotated variants (which after running VEP + inserting means were skipped)
         for v in get_variants_qs_for_annotation(annotation_version,
                                                 pipeline_type=annotation_run.pipeline_type,
                                                 min_variant_id=annotation_run.annotation_range_lock.min_variant_id,
@@ -92,6 +93,8 @@ def handle_vep_warnings(annotation_run: AnnotationRun, bulk_inserter):
                 reason = VEPSkippedReason.UNKNOWN_CONTIG
             elif v.pk in incomplete_variant_ids:
                 reason = VEPSkippedReason.INCOMPLETE
+            elif v.length > settings.ANNOTATION_VEP_SV_MAX_SIZE:
+                reason = VEPSkippedReason.TOO_LONG
             else:
                 reason = VEPSkippedReason.UNKNOWN
 

annotation/vep_annotation.py

@@ -162,6 +162,11 @@ def get_vep_command(vcf_filename, output_filename, genome_build: GenomeBuild, an
     if settings.ANNOTATION_VEP_DISTANCE is not None:
         cmd.extend(["--distance", str(settings.ANNOTATION_VEP_DISTANCE)])
 
+    if max_sv_size := settings.ANNOTATION_VEP_SV_MAX_SIZE:
+        vep_default_max_sv_size = 10_000_000
+        if max_sv_size != vep_default_max_sv_size:
+            cmd.extend(["--max_sv_size", str(max_sv_size)])
+
     if annotation_consortium == AnnotationConsortium.REFSEQ:
         cmd.append("--refseq")
 

variantgrid/settings/components/annotation_settings.py

@@ -37,6 +37,7 @@
 ANNOTATION_VEP_SV_OVERLAP_SAME_TYPE = True  # Only 'dup' for dups, false is all SVs overlap
 ANNOTATION_VEP_SV_OVERLAP_SINGLE_VALUE_METHOD = "lowest_af"  # "greatest_overlap", "lowest_af", "exact_or_lowest_af"
 ANNOTATION_VEP_SV_OVERLAP_MIN_FRACTION = 0.8
+ANNOTATION_VEP_SV_MAX_SIZE = 10_000_000  # VEP default = 10M
 
 ANNOTATION_MAX_BENIGN_RANKSCORE = 0.15
 ANNOTATION_MIN_PATHOGENIC_RANKSCORE = 0.85