Merge pull request #1052 from broadinstitute/sv-locus-alleles

Author: jklugherz

v03_pipeline/lib/annotations/sv.py

@@ -180,6 +180,7 @@ def gnomad_svs(
     gnomad_svs_ht: hl.Table,
     **_: Any,
 ) -> hl.Expression:
+    gnomad_svs_ht = gnomad_svs_ht.drop('locus', 'alleles')
     return gnomad_svs_ht.annotate(
         ID=gnomad_svs_ht.KEY,
     )[ht['info.GNOMAD_V4.1_TRUTH_VID']]

v03_pipeline/lib/reference_datasets/gnomad_svs.py

@@ -14,5 +14,4 @@ def get_ht(path: str, reference_genome: ReferenceGenome) -> hl.Table:
         N_HET=ht.info.N_HET,
         N_HOMREF=ht.info.N_HOMREF,
     )
-    ht = ht.key_by('KEY')
-    return ht.drop('locus', 'alleles')
+    return ht.key_by('KEY')

v03_pipeline/lib/reference_datasets/gnomad_svs_test.py

@@ -0,0 +1,93 @@
+import unittest
+from unittest.mock import patch
+
+import hail as hl
+
+from v03_pipeline.lib.model import ReferenceGenome
+from v03_pipeline.lib.reference_datasets.reference_dataset import ReferenceDataset
+
+TEST_GNOMAD_SVS_RAW_HT = (
+    'v03_pipeline/var/test/reference_datasets/raw/gnomad_svs_from_vcf.ht'
+)
+
+
+class GnomadSVsTest(unittest.TestCase):
+    @patch('v03_pipeline.lib.reference_datasets.gnomad_svs.vcf_to_ht')
+    def test_gnomad_svs(self, mock_vcf_to_ht):
+        mock_vcf_to_ht.return_value = hl.read_table(TEST_GNOMAD_SVS_RAW_HT)
+        ht = ReferenceDataset.gnomad_svs.get_ht(ReferenceGenome.GRCh38)
+        self.assertEqual(
+            ht.collect(),
+            [
+                hl.Struct(
+                    KEY='gnomAD-SV_v3_BND_chr1_1a45f73a',
+                    locus=hl.Locus(
+                        contig='chr1',
+                        position=10434,
+                        reference_genome=ReferenceGenome.GRCh38,
+                    ),
+                    alleles=['N', '<BND>'],
+                    AF=0.11413399875164032,
+                    AC=8474,
+                    AN=74246,
+                    N_HET=8426,
+                    N_HOMREF=28673,
+                ),
+                hl.Struct(
+                    KEY='gnomAD-SV_v3_BND_chr1_3fa36917',
+                    locus=hl.Locus(
+                        contig='chr1',
+                        position=10440,
+                        reference_genome=ReferenceGenome.GRCh38,
+                    ),
+                    alleles=['N', '<BND>'],
+                    AF=0.004201000090688467,
+                    AC=466,
+                    AN=110936,
+                    N_HET=466,
+                    N_HOMREF=55002,
+                ),
+                hl.Struct(
+                    KEY='gnomAD-SV_v3_BND_chr1_7bbf34b5',
+                    locus=hl.Locus(
+                        contig='chr1',
+                        position=10464,
+                        reference_genome=ReferenceGenome.GRCh38,
+                    ),
+                    alleles=['N', '<BND>'],
+                    AF=0.03698499873280525,
+                    AC=3119,
+                    AN=84332,
+                    N_HET=3115,
+                    N_HOMREF=39049,
+                ),
+                hl.Struct(
+                    KEY='gnomAD-SV_v3_BND_chr1_933a2971',
+                    locus=hl.Locus(
+                        contig='chr1',
+                        position=10450,
+                        reference_genome=ReferenceGenome.GRCh38,
+                    ),
+                    alleles=['N', '<BND>'],
+                    AF=0.3238990008831024,
+                    AC=21766,
+                    AN=67200,
+                    N_HET=21616,
+                    N_HOMREF=11909,
+                ),
+                hl.Struct(
+                    KEY='gnomAD-SV_v3_DUP_chr1_01c2781c',
+                    locus=hl.Locus(
+                        contig='chr1',
+                        position=10000,
+                        reference_genome=ReferenceGenome.GRCh38,
+                    ),
+                    alleles=['N', '<DUP>'],
+                    AF=0.0019970000721514225,
+                    AC=139,
+                    AN=69594,
+                    N_HET=139,
+                    N_HOMREF=34658,
+                ),
+            ],
+        )

v03_pipeline/var/test/reference_datasets/GRCh38/gnomad_svs/1.0.ht/.README.txt.crc

@@ -1,3 +1,3 @@
 This folder comprises a Hail (www.hail.is) native Table or MatrixTable.
   Written with version 0.2.133-4c60fddb171a
-  Created at 2025/02/16 18:40:38
+  Created at 2025/03/05 12:27:53

v03_pipeline/var/test/reference_datasets/GRCh38/gnomad_svs/1.0.ht/.metadata.json.gz.crc

@@ -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 2025/03/04 14:19:46