fix: gnomad_re should accept all nucleotide characters for ref/alt (#276

) (#277) Co-authored-by: Alex H. Wagner, PhD <[email protected]>
ga4gh · Nov 8, 2023 · 152d243 · 152d243
1 parent 4006ae2
commit 152d243
Show file tree

Hide file tree

Showing 4 changed files with 141 additions and 29 deletions.
diff --git a/src/ga4gh/vrs/extras/translator.py b/src/ga4gh/vrs/extras/translator.py
@@ -41,7 +41,10 @@ class Translator:
     """
 
     beacon_re = re.compile(r"(?P<chr>[^-]+)\s*:\s*(?P<pos>\d+)\s*(?P<ref>\w+)\s*>\s*(?P<alt>\w+)")
-    gnomad_re = re.compile(r"(?P<chr>[^-]+)-(?P<pos>\d+)-(?P<ref>[ACGTN]+)-(?P<alt>[ACGTN]+|\*|\.)", re.IGNORECASE)
+    gnomad_re = re.compile(
+        r"(?P<chr>[^-]+)-(?P<pos>\d+)-(?P<ref>[ACGTURYKMSWBDHVN]+)-(?P<alt>[ACGTURYKMSWBDHVN]+)",
+        re.IGNORECASE
+    )
     hgvs_re = re.compile(r"[^:]+:[cgnpr]\.")
     spdi_re = re.compile(r"(?P<ac>[^:]+):(?P<pos>\d+):(?P<del_len_or_seq>\w*):(?P<ins_seq>\w*)")
 

diff --git a/tests/extras/cassettes/test_from_gnomad.yaml b/tests/extras/cassettes/test_from_gnomad.yaml
@@ -9,7 +9,7 @@ interactions:
       Connection:
       - keep-alive
       User-Agent:
-      - python-requests/2.28.1
+      - python-requests/2.31.0
     method: GET
     uri: http://localhost:5000/seqrepo/1/sequence/GRCh38:19?start=44908821&end=44908822
   response:
@@ -23,7 +23,7 @@ interactions:
       Content-Type:
       - text/plain; charset=utf-8
       Date:
-      - Mon, 16 Jan 2023 16:32:42 GMT
+      - Tue, 07 Nov 2023 23:05:59 GMT
       Server:
       - Werkzeug/2.2.2 Python/3.10.4
     status:
@@ -39,33 +39,134 @@ interactions:
       Connection:
       - keep-alive
       User-Agent:
-      - python-requests/2.28.1
+      - python-requests/2.31.0
     method: GET
-    uri: http://localhost:5000/seqrepo/1/metadata/GRCh38:19
+    uri: http://localhost:5000/seqrepo/1/sequence/GRCh38:17?start=83129586&end=83129598
   response:
     body:
-      string: "{\n  \"added\": \"2016-08-24T08:19:02Z\",\n  \"aliases\": [\n    \"Ensembl:19\",\n
-        \   \"ensembl:19\",\n    \"GRCh38:19\",\n    \"GRCh38:chr19\",\n    \"GRCh38.p1:19\",\n
-        \   \"GRCh38.p1:chr19\",\n    \"GRCh38.p10:19\",\n    \"GRCh38.p10:chr19\",\n
-        \   \"GRCh38.p11:19\",\n    \"GRCh38.p11:chr19\",\n    \"GRCh38.p12:19\",\n
-        \   \"GRCh38.p12:chr19\",\n    \"GRCh38.p2:19\",\n    \"GRCh38.p2:chr19\",\n
-        \   \"GRCh38.p3:19\",\n    \"GRCh38.p3:chr19\",\n    \"GRCh38.p4:19\",\n    \"GRCh38.p4:chr19\",\n
-        \   \"GRCh38.p5:19\",\n    \"GRCh38.p5:chr19\",\n    \"GRCh38.p6:19\",\n    \"GRCh38.p6:chr19\",\n
-        \   \"GRCh38.p7:19\",\n    \"GRCh38.p7:chr19\",\n    \"GRCh38.p8:19\",\n    \"GRCh38.p8:chr19\",\n
-        \   \"GRCh38.p9:19\",\n    \"GRCh38.p9:chr19\",\n    \"MD5:b0eba2c7bb5c953d1e06a508b5e487de\",\n
-        \   \"NCBI:NC_000019.10\",\n    \"refseq:NC_000019.10\",\n    \"SEGUID:AHxM5/L8jIX08UhBBkKXkiO5rhY\",\n
-        \   \"SHA1:007c4ce7f2fc8c85f4f148410642979223b9ae16\",\n    \"VMC:GS_IIB53T8CNeJJdUqzn9V_JnRtQadwWCbl\",\n
-        \   \"sha512t24u:IIB53T8CNeJJdUqzn9V_JnRtQadwWCbl\",\n    \"ga4gh:SQ.IIB53T8CNeJJdUqzn9V_JnRtQadwWCbl\"\n
-        \ ],\n  \"alphabet\": \"ACGNT\",\n  \"length\": 58617616\n}\n"
+      string: GTTGWCACATGA
     headers:
       Connection:
       - close
       Content-Length:
-      - '1035'
+      - '12'
+      Content-Type:
+      - text/plain; charset=utf-8
+      Date:
+      - Tue, 07 Nov 2023 23:05:59 GMT
+      Server:
+      - Werkzeug/2.2.2 Python/3.10.4
+    status:
+      code: 200
+      message: OK
+- request:
+    body: null
+    headers:
+      Accept:
+      - '*/*'
+      Accept-Encoding:
+      - gzip, deflate
+      Connection:
+      - keep-alive
+      User-Agent:
+      - python-requests/2.31.0
+    method: GET
+    uri: http://localhost:5000/seqrepo/1/metadata/GRCh38:17
+  response:
+    body:
+      string: "{\n  \"added\": \"2016-08-27T23:52:54Z\",\n  \"aliases\": [\n    \"GRCh38:17\",\n
+        \   \"GRCh38:chr17\",\n    \"GRCh38.p1:17\",\n    \"GRCh38.p1:chr17\",\n    \"GRCh38.p10:17\",\n
+        \   \"GRCh38.p10:chr17\",\n    \"GRCh38.p11:17\",\n    \"GRCh38.p11:chr17\",\n
+        \   \"GRCh38.p12:17\",\n    \"GRCh38.p12:chr17\",\n    \"GRCh38.p2:17\",\n
+        \   \"GRCh38.p2:chr17\",\n    \"GRCh38.p3:17\",\n    \"GRCh38.p3:chr17\",\n
+        \   \"GRCh38.p4:17\",\n    \"GRCh38.p4:chr17\",\n    \"GRCh38.p5:17\",\n    \"GRCh38.p5:chr17\",\n
+        \   \"GRCh38.p6:17\",\n    \"GRCh38.p6:chr17\",\n    \"GRCh38.p7:17\",\n    \"GRCh38.p7:chr17\",\n
+        \   \"GRCh38.p8:17\",\n    \"GRCh38.p8:chr17\",\n    \"GRCh38.p9:17\",\n    \"GRCh38.p9:chr17\",\n
+        \   \"MD5:f9a0fb01553adb183568e3eb9d8626db\",\n    \"NCBI:NC_000017.11\",\n
+        \   \"refseq:NC_000017.11\",\n    \"SEGUID:s2Skupj8o6wdjf0aPrgOipAr67Q\",\n
+        \   \"SHA1:b364a4ba98fca3ac1d8dfd1a3eb80e8a902bebb4\",\n    \"VMC:GS_dLZ15tNO1Ur0IcGjwc3Sdi_0A6Yf4zm7\",\n
+        \   \"sha512t24u:dLZ15tNO1Ur0IcGjwc3Sdi_0A6Yf4zm7\",\n    \"ga4gh:SQ.dLZ15tNO1Ur0IcGjwc3Sdi_0A6Yf4zm7\"\n
+        \ ],\n  \"alphabet\": \"ACGKNRSTWY\",\n  \"length\": 83257441\n}\n"
+    headers:
+      Connection:
+      - close
+      Content-Length:
+      - '1004'
+      Content-Type:
+      - application/json
+      Date:
+      - Tue, 07 Nov 2023 23:05:59 GMT
+      Server:
+      - Werkzeug/2.2.2 Python/3.10.4
+    status:
+      code: 200
+      message: OK
+- request:
+    body: null
+    headers:
+      Accept:
+      - '*/*'
+      Accept-Encoding:
+      - gzip, deflate
+      Connection:
+      - keep-alive
+      User-Agent:
+      - python-requests/2.31.0
+    method: GET
+    uri: http://localhost:5000/seqrepo/1/sequence/GRCh38:7?start=1&end=17
+  response:
+    body:
+      string: NNNNNNNNNNNNNNNN
+    headers:
+      Connection:
+      - close
+      Content-Length:
+      - '16'
+      Content-Type:
+      - text/plain; charset=utf-8
+      Date:
+      - Tue, 07 Nov 2023 23:05:59 GMT
+      Server:
+      - Werkzeug/2.2.2 Python/3.10.4
+    status:
+      code: 200
+      message: OK
+- request:
+    body: null
+    headers:
+      Accept:
+      - '*/*'
+      Accept-Encoding:
+      - gzip, deflate
+      Connection:
+      - keep-alive
+      User-Agent:
+      - python-requests/2.31.0
+    method: GET
+    uri: http://localhost:5000/seqrepo/1/metadata/GRCh38:7
+  response:
+    body:
+      string: "{\n  \"added\": \"2016-08-27T21:23:35Z\",\n  \"aliases\": [\n    \"GRCh38:7\",\n
+        \   \"GRCh38:chr7\",\n    \"GRCh38.p1:7\",\n    \"GRCh38.p1:chr7\",\n    \"GRCh38.p10:7\",\n
+        \   \"GRCh38.p10:chr7\",\n    \"GRCh38.p11:7\",\n    \"GRCh38.p11:chr7\",\n
+        \   \"GRCh38.p12:7\",\n    \"GRCh38.p12:chr7\",\n    \"GRCh38.p2:7\",\n    \"GRCh38.p2:chr7\",\n
+        \   \"GRCh38.p3:7\",\n    \"GRCh38.p3:chr7\",\n    \"GRCh38.p4:7\",\n    \"GRCh38.p4:chr7\",\n
+        \   \"GRCh38.p5:7\",\n    \"GRCh38.p5:chr7\",\n    \"GRCh38.p6:7\",\n    \"GRCh38.p6:chr7\",\n
+        \   \"GRCh38.p7:7\",\n    \"GRCh38.p7:chr7\",\n    \"GRCh38.p8:7\",\n    \"GRCh38.p8:chr7\",\n
+        \   \"GRCh38.p9:7\",\n    \"GRCh38.p9:chr7\",\n    \"MD5:cc044cc2256a1141212660fb07b6171e\",\n
+        \   \"NCBI:NC_000007.14\",\n    \"refseq:NC_000007.14\",\n    \"SEGUID:4+JjCcBVhPCr8vdIhUKFycPv8bY\",\n
+        \   \"SHA1:e3e26309c05584f0abf2f748854285c9c3eff1b6\",\n    \"VMC:GS_F-LrLMe1SRpfUZHkQmvkVKFEGaoDeHul\",\n
+        \   \"sha512t24u:F-LrLMe1SRpfUZHkQmvkVKFEGaoDeHul\",\n    \"ga4gh:SQ.F-LrLMe1SRpfUZHkQmvkVKFEGaoDeHul\"\n
+        \ ],\n  \"alphabet\": \"ACGNRSTY\",\n  \"length\": 159345973\n}\n"
+    headers:
+      Connection:
+      - close
+      Content-Length:
+      - '977'
       Content-Type:
       - application/json
       Date:
-      - Mon, 16 Jan 2023 16:32:42 GMT
+      - Tue, 07 Nov 2023 23:05:59 GMT
       Server:
       - Werkzeug/2.2.2 Python/3.10.4
     status:
@@ -81,7 +182,7 @@ interactions:
       Connection:
       - keep-alive
       User-Agent:
-      - python-requests/2.28.1
+      - python-requests/2.31.0
     method: GET
     uri: http://localhost:5000/seqrepo/1/sequence/GRCh38:13?start=32936731&end=32936732
   response:
@@ -95,7 +196,7 @@ interactions:
       Content-Type:
       - text/plain; charset=utf-8
       Date:
-      - Mon, 16 Jan 2023 16:32:42 GMT
+      - Tue, 07 Nov 2023 23:05:59 GMT
       Server:
       - Werkzeug/2.2.2 Python/3.10.4
     status:
@@ -111,7 +212,7 @@ interactions:
       Connection:
       - keep-alive
       User-Agent:
-      - python-requests/2.28.1
+      - python-requests/2.31.0
     method: GET
     uri: http://localhost:5000/seqrepo/1/sequence/GRCh38:13?start=32936731&end=32936732
   response:
@@ -125,7 +226,7 @@ interactions:
       Content-Type:
       - text/plain; charset=utf-8
       Date:
-      - Mon, 16 Jan 2023 16:32:42 GMT
+      - Tue, 07 Nov 2023 23:05:59 GMT
       Server:
       - Werkzeug/2.2.2 Python/3.10.4
     status:
@@ -141,7 +242,7 @@ interactions:
       Connection:
       - keep-alive
       User-Agent:
-      - python-requests/2.28.1
+      - python-requests/2.31.0
     method: GET
     uri: http://localhost:5000/seqrepo/1/sequence/GRCh38:13?start=32936731&end=32936732
   response:
@@ -155,7 +256,7 @@ interactions:
       Content-Type:
       - text/plain; charset=utf-8
       Date:
-      - Mon, 16 Jan 2023 16:32:42 GMT
+      - Tue, 07 Nov 2023 23:05:59 GMT
       Server:
       - Werkzeug/2.2.2 Python/3.10.4
     status:
@@ -171,7 +272,7 @@ interactions:
       Connection:
       - keep-alive
       User-Agent:
-      - python-requests/2.28.1
+      - python-requests/2.31.0
     method: GET
     uri: http://localhost:5000/seqrepo/1/metadata/GRCh38:13
   response:
@@ -197,7 +298,7 @@ interactions:
       Content-Type:
       - application/json
       Date:
-      - Mon, 16 Jan 2023 16:32:42 GMT
+      - Tue, 07 Nov 2023 23:05:59 GMT
       Server:
       - Werkzeug/2.2.2 Python/3.10.4
     status:

diff --git a/tests/extras/test_translator.py b/tests/extras/test_translator.py
@@ -106,6 +106,14 @@ def test_from_beacon(tlr):
 def test_from_gnomad(tlr):
     assert tlr._from_gnomad(snv_inputs["gnomad"]).as_dict() == snv_output
 
+    assert tlr._from_gnomad("17-83129587-GTTGWCACATGA-G")
+
+    # Test valid characters
+    assert tlr._from_gnomad(
+        "7-2-ACGTURYKMSWBDHVN-ACGTURYKMSWBDHVN",
+        require_validation=False
+    )
+
     # Invalid input. Ref does not match regex
     assert not tlr._from_gnomad("13-32936732-helloworld-C")
 

diff --git a/tests/extras/test_vcf_annotation.py b/tests/extras/test_vcf_annotation.py
@@ -87,4 +87,4 @@ def test_get_vrs_object_invalid_input(vcf_annotator, caplog):
 
     # No ALT
     vcf_annotator._get_vrs_object("7-140753336-A-.", {}, [], "GRCh38")
-    assert "ValidationError when translating 7-140753336-A-. from gnomad" in caplog.text
+    assert "None was returned when translating 7-140753336-A-. from gnomad" in caplog.text