feat: add gnomad splice ai score

v03_pipeline/lib/reference_datasets/gnomad_exomes_test.py

@@ -67,6 +67,7 @@ def test_gnomad_exomes_38(self):
                         AF_POPMAX_OR_GLOBAL=0.009803921915590763,
                         FAF_AF=0.0,
                         Hemi=0,
+                        spliceai_ds_max=None,
                     ),
                 ],
             )

v03_pipeline/lib/reference_datasets/gnomad_genomes_test.py

@@ -67,6 +67,7 @@ def test_gnomad_genomes_38(self):
                         AF_POPMAX_OR_GLOBAL=3.779861071961932e-05,
                         FAF_AF=7.019999884505523e-06,
                         Hemi=0,
+                        spliceai_ds_max=None,
                     ),
                 ],
             )

v03_pipeline/lib/reference_datasets/gnomad_utils.py

@@ -51,5 +51,8 @@ def get_ht(
             0,
             ht.freq[ht.globals.freq_index_dict[hemi_field(reference_genome)]].AC,
         ),
+        **{'spliceai_ds_max': ht.in_silico_predictors.spliceai_ds_max}
+        if hasattr(ht, 'in_silico_predictors')
+        else {},
     )
     return ht.select_globals()

v03_pipeline/lib/reference_datasets/reference_dataset.py

@@ -375,7 +375,7 @@ def get_ht(
         },
         ReferenceGenome.GRCh38: {
             DATASET_TYPES: frozenset([DatasetType.SNV_INDEL]),
-            VERSION: '1.0',
+            VERSION: '1.1',
             PATH: 'gs://gcp-public-data--gnomad/release/4.1/ht/exomes/gnomad.exomes.v4.1.sites.ht',
         },
     },
@@ -387,7 +387,7 @@ def get_ht(
         },
         ReferenceGenome.GRCh38: {
             DATASET_TYPES: frozenset([DatasetType.SNV_INDEL]),
-            VERSION: '1.0',
+            VERSION: '1.1',
             PATH: 'gs://gcp-public-data--gnomad/release/4.1/ht/genomes/gnomad.genomes.v4.1.sites.ht',
         },
     },

v03_pipeline/lib/tasks/reference_data/update_variant_annotations_table_with_updated_reference_dataset_test.py

@@ -154,8 +154,8 @@ def test_update_vat_snv_indel_38(
                             splice_ai='1.1',
                             topmed='1.1',
                             hgmd='1.0',
-                            gnomad_exomes='1.0',
-                            gnomad_genomes='1.0',
+                            gnomad_exomes='1.1',
+                            gnomad_genomes='1.1',
                             screen='1.0',
                             gnomad_non_coding_constraint='1.0',
                         ),
@@ -234,6 +234,7 @@ def test_update_vat_snv_indel_38(
                             AF_POPMAX_OR_GLOBAL=0.0001119549197028391,
                             FAF_AF=9.315000352216884e-05,
                             Hemi=0,
+                            spliceai_ds_max=None,
                         ),
                         gnomad_genomes=hl.Struct(
                             AC=None,
@@ -243,6 +244,7 @@ def test_update_vat_snv_indel_38(
                             AF_POPMAX_OR_GLOBAL=None,
                             FAF_AF=None,
                             Hemi=None,
+                            spliceai_ds_max=None,
                         ),
                         gnomad_non_coding_constraint=hl.Struct(z_score=0.75),
                         screen=hl.Struct(region_type_ids=[1]),

v03_pipeline/lib/tasks/reference_data/updated_reference_dataset_query_test.py

@@ -177,6 +177,6 @@ def test_updated_query_high_af_variants(self) -> None:
         high_af_variants_ht = hl.read_table(high_af_variants_ht_path)
         self.assertEqual(
             hl.eval(high_af_variants_ht.version),
-            '1.0',
+            '1.1',
         )
         self.assertTrue(hasattr(high_af_variants_ht, 'is_gt_1_percent'))

v03_pipeline/lib/tasks/update_variant_annotations_table_with_new_samples_test.py

@@ -266,7 +266,7 @@ def test_multiple_update_vat(
             key='locus',
             globals=hl.Struct(
                 versions=hl.Struct(
-                    gnomad_genomes='1.0',
+                    gnomad_genomes='1.1',
                 ),
                 enums=hl.Struct(
                     gnomad_genomes=hl.Struct(),
@@ -499,8 +499,8 @@ def test_multiple_update_vat(
                             dbnsfp='1.0',
                             eigen='1.1',
                             exac='1.1',
-                            gnomad_exomes='1.0',
-                            gnomad_genomes='1.0',
+                            gnomad_exomes='1.1',
+                            gnomad_genomes='1.1',
                             splice_ai='1.1',
                             topmed='1.1',
                             gnomad_non_coding_constraint='1.0',
@@ -764,8 +764,8 @@ def test_update_vat_without_accessing_private_datasets(
                         dbnsfp='1.0',
                         eigen='1.1',
                         exac='1.1',
-                        gnomad_exomes='1.0',
-                        gnomad_genomes='1.0',
+                        gnomad_exomes='1.1',
+                        gnomad_genomes='1.1',
                         splice_ai='1.1',
                         topmed='1.1',
                         gnomad_non_coding_constraint='1.0',

v03_pipeline/var/test/reference_datasets/GRCh38/gnomad_coding_and_noncoding/1.1.ht/README.txt

@@ -0,0 +1,3 @@
+This folder comprises a Hail (www.hail.is) native Table or MatrixTable.
+  Written with version 0.2.133-4c60fddb171a
+  Created at 2025/06/03 20:30:00

v03_pipeline/var/test/reference_datasets/GRCh38/gnomad_exomes/1.1.ht/README.txt

@@ -0,0 +1,3 @@
+This folder comprises a Hail (www.hail.is) native Table or MatrixTable.
+  Written with version 0.2.133-4c60fddb171a
+  Created at 2025/06/03 17:48:37

v03_pipeline/var/test/reference_datasets/GRCh38/gnomad_genomes/1.1.ht/README.txt

@@ -0,0 +1,3 @@
+This folder comprises a Hail (www.hail.is) native Table or MatrixTable.
+  Written with version 0.2.133-4c60fddb171a
+  Created at 2025/06/03 17:50:03