wip: rm gnomad vcf deletions

consider them delins

tests/conftest.py

@@ -401,6 +401,34 @@ def genomic_del1_seq_loc():
  "type": "SequenceLocation",
  }
 
+@pytest.fixture(scope="session")
+def genomic_del1():
+ """Create test fixture containing params for genomic del VD."""
+ params = {
+ "id": "normalize.variation:NC_000003.12%3Ag.10149811del",
+ "type": "VariationDescriptor",
+ "variation_id": "",
+ "variation": dict(),
+ "molecule_context": "genomic",
+ "structural_type": "SO:0000159",
+ "vrs_ref_allele_seq": "T",
+ }
+ return params
+
+
+@pytest.fixture(scope="session")
+def genomic_del1_lse(genomic_del1, genomic_del1_seq_loc):
+ """Create a test fixture for genomic del LSE."""
+ _id = "ga4gh:VA.jUeT1n4AuBzwtt5TT-Iaac1KasATWjKE"
+ genomic_del1["variation_id"] = _id
+ genomic_del1["variation"] = {
+ "type": "Allele",
+ "_id": _id,
+ "location": genomic_del1_seq_loc,
+ "state": {"type": "LiteralSequenceExpression", "sequence": ""},
+ }
+ return VariationDescriptor(**genomic_del1)
+
 
 @pytest.fixture(scope="session")
 def genomic_del1_38_cn(genomic_del1_seq_loc):
@@ -428,6 +456,35 @@ def genomic_del2_seq_loc():
  }
 
 
+@pytest.fixture(scope="session")
+def genomic_del2():
+ """Create test fixture containing params for genomic del VD."""
+ params = {
+ "id": "normalize.variation:NC_000003.12%3Ag.10146595_10146613del",
+ "type": "VariationDescriptor",
+ "variation_id": "",
+ "variation": dict(),
+ "molecule_context": "genomic",
+ "structural_type": "SO:0000159",
+ "vrs_ref_allele_seq": "ATGTTGACGGACAGCCTAT",
+ }
+ return params
+
+
+@pytest.fixture(scope="session")
+def genomic_del2_lse(genomic_del2, genomic_del2_seq_loc):
+ """Create a test fixture for genomic del LSE."""
+ _id = "ga4gh:VA.CSWNhR5w_geMmJTxkbO3UCLCvT0S2Ypx"
+ genomic_del2["variation_id"] = _id
+ genomic_del2["variation"] = {
+ "type": "Allele",
+ "_id": _id,
+ "location": genomic_del2_seq_loc,
+ "state": {"type": "LiteralSequenceExpression", "sequence": ""},
+ }
+ return VariationDescriptor(**genomic_del2)
+
+
 @pytest.fixture(scope="session")
 def genomic_del2_38_cn(genomic_del2_seq_loc):
  """Create test fixture for copy number count del1 on GRCh38"""

tests/fixtures/classifiers.yml

@@ -108,6 +108,9 @@ genomic_delins:
  - query: 3-37050340-AAAAGCTTTA-GAGGCTTT
  - query: 16-68846036-AG-TGAGTTT
  - query: X-70350063-AG-AGGCAGCGCATAAAGCGCATTCTCCG
+ - query: Y-1313-ATTGAC-a
+ - query: chr1-2-ca-C
+ - query: 1-55509715-AC-A
  should_not_match:
  - query: N_000017.10:c.1423_1424delinsGT
  - query: g.1423delinsX
@@ -151,9 +154,6 @@ genomic_deletion:
  - query: NC_000004.11:g.55593610_55593615delTTGTTG
  - query: NC_000003.11:g.10183645del
  - query: NC_000003.11:g.10188302delG
- - query: Y-1313-ATTGAC-a
- - query: chr1-2-ca-C
- - query: 1-55509715-AC-A
  should_not_match:
  - query: GENE g.152419920_152419921delinsAG
  - query: GENE g.152419920_152419921delAinsG

tests/fixtures/translators.yml

@@ -775,6 +775,33 @@ genomic_delins:
  "type": "Allele"
  }
  ]
+ - query: 17-7578455-CGCGG-CGCG
+ variations: [
+ {
+ "_id": "ga4gh:VA.BFl5Ja-FnzlReimbu_eRTfcyvXHZbjUC",
+ "type": "Allele",
+ "location": {
+ "_id": "ga4gh:VSL.KEPjHEPUVfIwk8R8xyluX8IW5-9pY6TE",
+ "type": "SequenceLocation",
+ "sequence_id": "ga4gh:SQ.AjWXsI7AkTK35XW9pgd3UbjpC3MAevlz",
+ "interval": {
+ "type": "SequenceInterval",
+ "start": {
+ "type": "Number",
+ "value": 7578457
+ },
+ "end": {
+ "type": "Number",
+ "value": 7578459
+ }
+ }
+ },
+ "state": {
+ "type": "LiteralSequenceExpression",
+ "sequence": "G"
+ }
+ }
+ ]
 
 protein_deletion:
  tests:

tests/fixtures/validators.yml

@@ -114,6 +114,8 @@ genomic_delins:
  - query: 3-37050340-AAAAGCTTTA-GAGGCTTT
  - query: 16-68846036-AG-TGAGTTT
  - query: X-70350063-AG-AGGCAGCGCATAAAGCGCATTCTCCG
+ - query: 16-2138199-GTGAG-G
+ - query: 1-55509715-AC-A
  should_not_match:
  - query: NC_000023.21:g.32386323delinsGA
  - query: NC_000007.13:g.159138664delinsAT
@@ -123,6 +125,8 @@ genomic_delins:
  - query: 3-37050340-AAAAGCGTTA-GAGGCTTT
  - query: 16-68846036-AC-TGAGTTT
  - query: X-70350063-CC-AGGCAGCGCATAAAGCGCATTCTCCG
+ - query: 1-55509715-TC-A
+ - query: 16-2138199-GTGAT-G
 
 protein_deletion:
  should_match:
@@ -160,13 +164,9 @@ genomic_deletion:
  - query: NC_000003.12:g.10146527_10146528del
  - query: NC_000003.11:g.10191495delT
  - query: VHL g.10188279_10188297del
- - query: 16-2138199-GTGAG-G
- - query: 1-55509715-AC-A
  should_not_match:
  - query: NC_000003.11:g.10191454654654654495delT
  - query: NC_000003.11:g.10188297_10188279del
- - query: 1-55509715-TC-A
- - query: 16-2138199-GTGAT-G
 
 protein_insertion:
  should_match:

tests/test_gnomad_vcf_to_protein.py

@@ -1,9 +1,12 @@
 """Module for testing gnomad_vcf_to_protein works correctly"""
+from copy import deepcopy
+
 import pytest
 from ga4gh.vrsatile.pydantic.vrsatile_models import VariationDescriptor
 
 from tests.conftest import assertion_checks
 from variation.gnomad_vcf_to_protein_variation import dna_to_rna
+from variation.schemas.classification_response_schema import SequenceOntology
 
 
 @pytest.fixture(scope="module")
@@ -347,16 +350,20 @@ async def test_insertion(test_handler, protein_insertion, protein_insertion2):
 async def test_deletion(test_handler, protein_deletion_np_range, cdk11a_e314del):
  """Test that deletion queries return correct response"""
  resp = await test_handler.gnomad_vcf_to_protein("17-39723966-TTGAGGGAAAACACAT-T")
+ expected = deepcopy(protein_deletion_np_range)
+ expected.structural_type = SequenceOntology.DELINS
  assertion_checks(
  resp.variation_descriptor,
- protein_deletion_np_range,
+ expected,
  "17-39723966-TTGAGGGAAAACACAT-T",
  ignore_id=True,
  )
  assert resp.warnings == []
 
  resp = await test_handler.gnomad_vcf_to_protein("1-1708855-TTCC-T")
- assertion_checks(resp.variation_descriptor, cdk11a_e314del, "1-1708855-TTCC-T")
+ expected = deepcopy(cdk11a_e314del)
+ expected.structural_type = SequenceOntology.DELINS
+ assertion_checks(resp.variation_descriptor, expected, "1-1708855-TTCC-T")
  assert resp.warnings == []
 
 

tests/test_hgvs_dup_del_mode.py

@@ -884,35 +884,6 @@ def genomic_dup6_free_text_rse_lse(genomic_dup6_free_text):
  return VariationDescriptor(**params)
 
 
-@pytest.fixture(scope="module")
-def genomic_del1():
- """Create test fixture containing params for genomic del VD."""
- params = {
- "id": "normalize.variation:NC_000003.12%3Ag.10149811del",
- "type": "VariationDescriptor",
- "variation_id": "",
- "variation": dict(),
- "molecule_context": "genomic",
- "structural_type": "SO:0000159",
- "vrs_ref_allele_seq": "T",
- }
- return params
-
-
-@pytest.fixture(scope="module")
-def genomic_del1_lse(genomic_del1, genomic_del1_seq_loc):
- """Create a test fixture for genomic del LSE."""
- _id = "ga4gh:VA.jUeT1n4AuBzwtt5TT-Iaac1KasATWjKE"
- genomic_del1["variation_id"] = _id
- genomic_del1["variation"] = {
- "type": "Allele",
- "_id": _id,
- "location": genomic_del1_seq_loc,
- "state": {"type": "LiteralSequenceExpression", "sequence": ""},
- }
- return VariationDescriptor(**genomic_del1)
-
-
 @pytest.fixture(scope="module")
 def genomic_del1_cn(genomic_del1, genomic_del1_38_cn):
  """Create a test fixture for genomic del copy number count."""
@@ -1036,36 +1007,6 @@ def genomic_del1_free_text_rse(genomic_del1_free_text, genomic_del1_free_text_se
  genomic_del1_free_text["variation_id"] = _id
  return VariationDescriptor(**genomic_del1_free_text)
 
-
-@pytest.fixture(scope="module")
-def genomic_del2():
- """Create test fixture containing params for genomic del VD."""
- params = {
- "id": "normalize.variation:NC_000003.12%3Ag.10146595_10146613del",
- "type": "VariationDescriptor",
- "variation_id": "",
- "variation": dict(),
- "molecule_context": "genomic",
- "structural_type": "SO:0000159",
- "vrs_ref_allele_seq": "ATGTTGACGGACAGCCTAT",
- }
- return params
-
-
-@pytest.fixture(scope="module")
-def genomic_del2_lse(genomic_del2, genomic_del2_seq_loc):
- """Create a test fixture for genomic del LSE."""
- _id = "ga4gh:VA.CSWNhR5w_geMmJTxkbO3UCLCvT0S2Ypx"
- genomic_del2["variation_id"] = _id
- genomic_del2["variation"] = {
- "type": "Allele",
- "_id": _id,
- "location": genomic_del2_seq_loc,
- "state": {"type": "LiteralSequenceExpression", "sequence": ""},
- }
- return VariationDescriptor(**genomic_del2)
-
-
 @pytest.fixture(scope="module")
 def genomic_del2_cn(genomic_del2, genomic_del2_38_cn):
  """Create a test fixture for genomic del copy number count."""
@@ -2419,14 +2360,6 @@ async def test_genomic_del1(
  resp.variation_descriptor, genomic_del1_free_text_lse, q, ignore_id=True
  )
 
- # gnomad vcf
- q = "3-10149810-CT-C" # 38
- resp = await test_handler.normalize(q)
- assertion_checks(resp.variation_descriptor, genomic_del1_lse, q, ignore_id=True)
-
- resp = await test_handler.normalize(q, HGVSDupDelModeOption.COPY_NUMBER_COUNT)
- assertion_checks(resp.variation_descriptor, genomic_del1_lse, q, ignore_id=True)
-
  # Invalid
  invalid_queries = [
  "NC_000003.11:g.198022431del",
@@ -2513,15 +2446,6 @@ async def test_genomic_del2(
  resp.variation_descriptor, genomic_del2_free_text_default, q, ignore_id=True
  )
 
- # gnomad vcf
- q = "3-10146594-AATGTTGACGGACAGCCTAT-A"
- resp = await test_handler.normalize(q, HGVSDupDelModeOption.DEFAULT)
- assertion_checks(resp.variation_descriptor, genomic_del2_lse, q, ignore_id=True)
-
- q = "3-10188278-AATGTTGACGGACAGCCTAT-A"
- resp = await test_handler.normalize(q)
- assertion_checks(resp.variation_descriptor, genomic_del2_lse, q, ignore_id=True)
-
  # Invalid
  invalid_queries = [
  "NC_000003.12:g.10146595_198295580del",