Get gnomADv2 new fields to be applied by VEP

annotation/migrations/0082_new_vep_110_columns_v3.py

@@ -6,6 +6,7 @@
 
 
 def _new_vep_110_annotation(apps, _schema_editor):
+    VEP_CUSTOM_GNOMAD_2 = 'g'
     VEP_CUSTOM_GNOMAD_3 = 'n'
     VEP_CUSTOM_GNOMAD_4 = 'o'
 
@@ -26,23 +27,23 @@ def _new_vep_110_annotation(apps, _schema_editor):
         {'column': 'gnomad2_ac', 'variant_grid_column_id': 'gnomad_ac',
          'genome_build_id': 'GRCh37', 'pipeline_type': 'S', 'category': 'F', 'source_field': 'AC',
          'source_field_processing_description': 'Sum of exome AC + genome AC',
-         'vep_custom': 'g', 'source_field_has_custom_prefix': True},
+         'vep_custom': VEP_CUSTOM_GNOMAD_2, 'source_field_has_custom_prefix': True},
         {'column': 'gnomad2_popmax_ac', 'variant_grid_column_id': 'gnomad_popmax_ac',
          'genome_build_id': 'GRCh37', 'pipeline_type': 'S', 'category': 'F', 'source_field': 'AC_popmax',
          'source_field_processing_description': 'Sum of exome AC_popmax + genome AC_popmax',
-         'vep_custom': 'g', 'source_field_has_custom_prefix': True},
+         'vep_custom': VEP_CUSTOM_GNOMAD_2, 'source_field_has_custom_prefix': True},
         {'column': 'gnomad2_an', 'variant_grid_column_id': 'gnomad_an',
          'genome_build_id': 'GRCh37', 'pipeline_type': 'S', 'category': 'F', 'source_field': 'AN',
          'source_field_processing_description': 'Sum of exome AN + genome AN',
-         'vep_custom': 'g', 'source_field_has_custom_prefix': True},
+         'vep_custom': VEP_CUSTOM_GNOMAD_2, 'source_field_has_custom_prefix': True},
         {'column': 'gnomad2_popmax_an', 'variant_grid_column_id': 'gnomad_popmax_an',
          'genome_build_id': 'GRCh37', 'pipeline_type': 'S', 'category': 'F', 'source_field': 'AN_popmax',
          'source_field_processing_description': 'Sum of exome AN_popmax + genome AN_popmax',
-         'vep_custom': 'g', 'source_field_has_custom_prefix': True},
+         'vep_custom': VEP_CUSTOM_GNOMAD_2, 'source_field_has_custom_prefix': True},
         {'column': 'gnomad2_nonpar', 'variant_grid_column_id': 'gnomad_non_par',
          'genome_build_id': 'GRCh37', 'pipeline_type': 'S', 'category': 'F', 'source_field': 'nonpar',
          'source_field_processing_description': 'nonpar from genomes',
-         'vep_custom': 'g', 'source_field_has_custom_prefix': True},
+         'vep_custom': VEP_CUSTOM_GNOMAD_2, 'source_field_has_custom_prefix': True},
 
         # gnomAD 4 - issue #938
         {'column': 'gnomad4_ac', 'source_field_has_custom_prefix': True, 'min_vep_columns_version': 3,

annotation/migrations/0086_gnomad2_xy.py

@@ -0,0 +1,33 @@
+# Generated by Django 4.1.4 on 2023-12-07 11:48
+
+from django.db import migrations
+
+from library.django_utils import bulk_insert_class_data
+
+
+def _gnomad2xy(apps, _schema_editor):
+    VEP_CUSTOM_GNOMAD_2 = 'g'
+    FREQUENCY_DATA = 'F'
+
+    COLUMN_VEP_FIELD = [
+        {'column': 'gnomad2_xy_ac', 'source_field_has_custom_prefix': True, 'min_vep_columns_version': 3,
+         'vep_custom': VEP_CUSTOM_GNOMAD_2, 'variant_grid_column_id': 'gnomad_xy_ac', 'source_field': 'AC_male',
+         'category': FREQUENCY_DATA, 'genome_build_id': 'GRCh37'},
+        {'column': 'gnomad2_xy_af', 'source_field_has_custom_prefix': True, 'min_vep_columns_version': 3,
+         'vep_custom': VEP_CUSTOM_GNOMAD_2, 'variant_grid_column_id': 'gnomad_xy_af', 'source_field': 'AF_male',
+         'category': FREQUENCY_DATA, 'genome_build_id': 'GRCh37'},
+        {'column': 'gnomad2_xy_an', 'source_field_has_custom_prefix': True, 'min_vep_columns_version': 3,
+         'vep_custom': VEP_CUSTOM_GNOMAD_2, 'variant_grid_column_id': 'gnomad_xy_an', 'source_field': 'AN_male',
+         'category': FREQUENCY_DATA, 'genome_build_id': 'GRCh37'},
+    ]
+    bulk_insert_class_data(apps, "annotation", [("ColumnVEPField", COLUMN_VEP_FIELD)])
+
+
+class Migration(migrations.Migration):
+    dependencies = [
+        ("annotation", "0085_alter_dbnsfpgeneannotationversion_options_and_more"),
+    ]
+
+    operations = [
+        migrations.RunPython(_gnomad2xy)
+    ]

variantgrid/settings/components/default_settings.py

@@ -271,7 +271,7 @@
             "dbscsnv": "annotation_data/GRCh38/dbscSNV1.1_GRCh38.txt.gz",
             "gnomad2": "annotation_data/GRCh38/gnomad2.1.1_GRCh38_combined_af.vcf.bgz",
             "gnomad3": "annotation_data/GRCh38/gnomad3.1_GRCh38_merged.vcf.bgz",
-            "gnomad4": "annotation_data/GRCh38/gnomad4.0_GRCh38_combined_af.vcf.bgz",
+            "gnomad4": "annotation_data/GRCh38/gnomad4.0.GRCh38.combined_af.vcf.bgz",
             "mastermind": "annotation_data/GRCh38/mastermind_cited_variants_reference-2022.04.02-grch38.vcf.gz",
             "mave": "annotation_data/GRCh38/MaveDB_variants.tsv.gz",
             "maxentscan": "annotation_data/all_builds/maxentscan",

variantgrid/settings/env/vgtest.py

@@ -28,21 +28,26 @@
 
 
 ANNOTATION[BUILD_GRCH37].update({
-    "columns_version": 2,
+    "columns_version": 3,
     "reference_fasta": os.path.join(_ANNOTATION_FASTA_BASE_DIR, "GCF_000001405.25_GRCh37.p13_genomic.fna.gz"),
 })
 
 ANNOTATION[BUILD_GRCH38].update({
-    "columns_version": 2,
+    "columns_version": 3,
     "enabled": True,
     "reference_fasta": os.path.join(_ANNOTATION_FASTA_BASE_DIR, "GCF_000001405.39_GRCh38.p13_genomic.fna.gz"),
 })
 
+
 ANNOTATION[BUILD_GRCH37]["vep_config"].update({
-    "dbnsfp": "annotation_data/GRCh37/dbNSFP4.3a.grch37.stripped.gz",
+    "cosmic": "annotation_data/GRCh37/Cosmic_GenomeScreensMutant_v99_GRCh37.vcf.gz",
+    "dbnsfp": "annotation_data/GRCh37/dbNSFP4.5a.grch37.stripped.gz",
+    "mastermind": "annotation_data/GRCh37/mastermind_cited_variants_reference-2023.10.02-grch37.vcf.gz",
 })
 ANNOTATION[BUILD_GRCH38]["vep_config"].update({
-    "dbnsfp": "annotation_data/GRCh38/dbNSFP4.3a.grch38.stripped.gz",
+    "cosmic": "annotation_data/GRCh38/Cosmic_GenomeScreensMutant_v99_GRCh38.vcf.gz",
+    "dbnsfp": "annotation_data/GRCh38/dbNSFP4.5a.grch38.stripped.gz",
+    "mastermind": "annotation_data/GRCh38/mastermind_cited_variants_reference-2023.10.02-grch38.vcf.gz",
 })
 
 ANNOTATION_VCF_DUMP_DIR = os.path.join(ANNOTATION_REFERENCE_BASE_DIR, 'scratch')