feat: cnv translator supports uncertain ranges for g. and m. (#387)

Added capability for CNV translator to support `Range` endpoints in SequenceLocations for g. and m. hgvs expressions.

setup.cfg

@@ -92,7 +92,7 @@ extras =
     psycopg2-binary
     biocommons.seqrepo>=0.5.1
     bioutils>=0.5.2
-    hgvs>=1.4
+    hgvs@git+https://github.com/biocommons/hgvs@225-uncertain-ranges
     requests
     dill~=0.3.7
     click

src/ga4gh/vrs/extras/translator.py

@@ -10,6 +10,7 @@
 from typing import Optional, Union
 from ga4gh.vrs.dataproxy import create_dataproxy, _DataProxy
 from ga4gh.vrs.extras.decorators import lazy_property
+from hgvs.location import Interval, SimplePosition, AAPosition, BaseOffsetInterval
 import logging
 import re
 
@@ -438,11 +439,14 @@ def _from_hgvs(self, hgvs_dup_del_expr: str, **kwargs):
             copy_change: Copy change. If not provided, default is efo:0030067 for
                 deletions and efo:0030070 for duplications
         """
-        # sv = self._get_parsed_hgvs(hgvs_dup_del_expr)
         sv = self.hgvs_tools.parse(hgvs_dup_del_expr)
         if not sv:
             return None
 
+        if sv.type not in {"g", "m"}:
+            err_msg = "Only 'g' and 'm' reference sequences are supported"
+            raise ValueError(err_msg)
+
         sv_type = self.hgvs_tools.get_edit_type(sv)
         if sv_type not in {"del", "dup"}:
             raise ValueError("Must provide a 'del' or 'dup'")
@@ -454,10 +458,13 @@ def _from_hgvs(self, hgvs_dup_del_expr: str, **kwargs):
         if not refget_accession:
             return None
 
+        start = self._get_vrs_loc_start_end_val(sv.posedit.pos.start, is_start=True)
+        end = self._get_vrs_loc_start_end_val(sv.posedit.pos.end, is_start=False)
+
         location = models.SequenceLocation(
             sequenceReference=models.SequenceReference(refgetAccession=refget_accession),
-            start=sv.posedit.pos.start.base - 1,
-            end=sv.posedit.pos.end.base
+            start=start,
+            end=end
         )
 
         copies = kwargs.get("copies")
@@ -472,6 +479,43 @@ def _from_hgvs(self, hgvs_dup_del_expr: str, **kwargs):
         cnv =self._post_process_imported_cnv(cnv)
         return cnv
 
+    @staticmethod
+    def _get_vrs_loc_start_end_val(
+        pos: Union[SimplePosition, AAPosition, BaseOffsetInterval, Interval],
+        is_start: bool = True
+    ) -> Union[int, models.Range]:
+        """Get VRS Location start or end value
+
+        :param pos: biocommons hgvs location instance for position
+        :param is_start: ``True`` if ``pos`` represents VRS Sequence Location start.
+            ``False`` if ``pos`` represents VRS Sequence Location end.
+        :raise ValueError: If unsupported biocommons hgvs location is passed
+        :return: VRS Location start or end value using inter-residue positions
+        """
+        def _get_pos_value(
+            position: Optional[int],
+            do_subtract_1: bool
+        ) -> Optional[int]:
+            """Get position value
+
+            :param position: Position
+            :param do_subtract_1: Whether or not we need to subtract 1 for ``position``
+            :return: Adjusted position value
+            """
+            return position - 1 if do_subtract_1 else position
+
+        if isinstance(pos, (SimplePosition, AAPosition, BaseOffsetInterval)):
+            vrs_loc_pos_val = _get_pos_value(pos.base, is_start)
+        elif isinstance(pos, Interval):
+            start_val = _get_pos_value(pos.start.base, is_start and pos.start.base is not None)
+            end_val = _get_pos_value(pos.end.base, is_start and pos.end.base is not None)
+            vrs_loc_pos_val = start_val if start_val == end_val else models.Range([start_val, end_val])
+        else:
+            err_msg = f"HGVS Location is not supported: {type(pos)}"
+            raise ValueError(err_msg)
+
+        return vrs_loc_pos_val
+
     def _post_process_imported_cnv(self, copy_number):
         """Provide common post-processing for imported Copy Numbers IN-PLACE."""
         if self.identify:

tests/extras/cassettes/test_from_hgvs_cx[NC_000005.9:g.(90136803)_(90159675)dup-None-expected6].yaml

@@ -0,0 +1,42 @@
+interactions:
+- 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/refseq:NC_000005.9
+  response:
+    body:
+      string: "{\n  \"added\": \"2016-08-24T05:17:01Z\",\n  \"aliases\": [\n    \"GRCh37:5\",\n
+        \   \"GRCh37:chr5\",\n    \"GRCh37.p10:5\",\n    \"GRCh37.p10:chr5\",\n    \"GRCh37.p11:5\",\n
+        \   \"GRCh37.p11:chr5\",\n    \"GRCh37.p12:5\",\n    \"GRCh37.p12:chr5\",\n
+        \   \"GRCh37.p13:5\",\n    \"GRCh37.p13:chr5\",\n    \"GRCh37.p2:5\",\n    \"GRCh37.p2:chr5\",\n
+        \   \"GRCh37.p5:5\",\n    \"GRCh37.p5:chr5\",\n    \"GRCh37.p9:5\",\n    \"GRCh37.p9:chr5\",\n
+        \   \"MD5:0740173db9ffd264d728f32784845cd7\",\n    \"NCBI:NC_000005.9\",\n
+        \   \"refseq:NC_000005.9\",\n    \"SEGUID:Ja+pA+dtRy6jSKdOZXN58wY0rK4\",\n
+        \   \"SHA1:25afa903e76d472ea348a74e657379f30634acae\",\n    \"VMC:GS_vbjOdMfHJvTjK_nqvFvpaSKhZillW0SX\",\n
+        \   \"sha512t24u:vbjOdMfHJvTjK_nqvFvpaSKhZillW0SX\",\n    \"ga4gh:SQ.vbjOdMfHJvTjK_nqvFvpaSKhZillW0SX\",\n
+        \   \"hs37-1kg:5\",\n    \"hs37d5:5\"\n  ],\n  \"alphabet\": \"ACGNT\",\n
+        \ \"length\": 180915260\n}\n"
+    headers:
+      Connection:
+      - close
+      Content-Length:
+      - '803'
+      Content-Type:
+      - application/json
+      Date:
+      - Fri, 05 Apr 2024 19:19:57 GMT
+      Server:
+      - Werkzeug/2.2.2 Python/3.10.4
+    status:
+      code: 200
+      message: OK
+version: 1

tests/extras/cassettes/test_from_hgvs_cx[NC_000009.11:g.(?_108337304)_(108337428_?)del-None-expected5].yaml

@@ -0,0 +1,42 @@
+interactions:
+- 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/refseq:NC_000009.11
+  response:
+    body:
+      string: "{\n  \"added\": \"2016-08-24T05:18:51Z\",\n  \"aliases\": [\n    \"GRCh37:9\",\n
+        \   \"GRCh37:chr9\",\n    \"GRCh37.p10:9\",\n    \"GRCh37.p10:chr9\",\n    \"GRCh37.p11:9\",\n
+        \   \"GRCh37.p11:chr9\",\n    \"GRCh37.p12:9\",\n    \"GRCh37.p12:chr9\",\n
+        \   \"GRCh37.p13:9\",\n    \"GRCh37.p13:chr9\",\n    \"GRCh37.p2:9\",\n    \"GRCh37.p2:chr9\",\n
+        \   \"GRCh37.p5:9\",\n    \"GRCh37.p5:chr9\",\n    \"GRCh37.p9:9\",\n    \"GRCh37.p9:chr9\",\n
+        \   \"MD5:3e273117f15e0a400f01055d9f393768\",\n    \"NCBI:NC_000009.11\",\n
+        \   \"refseq:NC_000009.11\",\n    \"SEGUID:06BhLSlH1xeVNSYu0zRK9qIVmFg\",\n
+        \   \"SHA1:d3a0612d2947d7179535262ed3344af6a2159858\",\n    \"VMC:GS_HBckYGQ4wYG9APHLpjoQ9UUe9v7NxExt\",\n
+        \   \"sha512t24u:HBckYGQ4wYG9APHLpjoQ9UUe9v7NxExt\",\n    \"ga4gh:SQ.HBckYGQ4wYG9APHLpjoQ9UUe9v7NxExt\",\n
+        \   \"hs37-1kg:9\",\n    \"hs37d5:9\"\n  ],\n  \"alphabet\": \"ACGNT\",\n
+        \ \"length\": 141213431\n}\n"
+    headers:
+      Connection:
+      - close
+      Content-Length:
+      - '805'
+      Content-Type:
+      - application/json
+      Date:
+      - Fri, 05 Apr 2024 19:12:51 GMT
+      Server:
+      - Werkzeug/2.2.2 Python/3.10.4
+    status:
+      code: 200
+      message: OK
+version: 1

tests/extras/cassettes/test_from_hgvs_cx[NC_000013.11:g.26440969_26443305del-CopyChange.EFO_0030069-expected0].yaml

@@ -35,7 +35,7 @@ interactions:
       Content-Type:
       - application/json
       Date:
-      - Tue, 26 Mar 2024 16:28:06 GMT
+      - Fri, 05 Apr 2024 19:12:51 GMT
       Server:
       - Werkzeug/2.2.2 Python/3.10.4
     status:

tests/extras/cassettes/test_from_hgvs_cx[NC_000023.11:g.(31060227_31100351)_(33274278_33417151)dup-None-expected4].yaml

@@ -0,0 +1,43 @@
+interactions:
+- 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/refseq:NC_000023.11
+  response:
+    body:
+      string: "{\n  \"added\": \"2016-08-27T23:57:18Z\",\n  \"aliases\": [\n    \"GRCh38:X\",\n
+        \   \"GRCh38:chrX\",\n    \"GRCh38.p1:X\",\n    \"GRCh38.p1:chrX\",\n    \"GRCh38.p10:X\",\n
+        \   \"GRCh38.p10:chrX\",\n    \"GRCh38.p11:X\",\n    \"GRCh38.p11:chrX\",\n
+        \   \"GRCh38.p12:X\",\n    \"GRCh38.p12:chrX\",\n    \"GRCh38.p2:X\",\n    \"GRCh38.p2:chrX\",\n
+        \   \"GRCh38.p3:X\",\n    \"GRCh38.p3:chrX\",\n    \"GRCh38.p4:X\",\n    \"GRCh38.p4:chrX\",\n
+        \   \"GRCh38.p5:X\",\n    \"GRCh38.p5:chrX\",\n    \"GRCh38.p6:X\",\n    \"GRCh38.p6:chrX\",\n
+        \   \"GRCh38.p7:X\",\n    \"GRCh38.p7:chrX\",\n    \"GRCh38.p8:X\",\n    \"GRCh38.p8:chrX\",\n
+        \   \"GRCh38.p9:X\",\n    \"GRCh38.p9:chrX\",\n    \"MD5:2b3a55ff7f58eb308420c8a9b11cac50\",\n
+        \   \"NCBI:NC_000023.11\",\n    \"refseq:NC_000023.11\",\n    \"SEGUID:Z9QbQrrPjpjXSMJesDYqC3A43lA\",\n
+        \   \"SHA1:67d41b42bacf8e98d748c25eb0362a0b7038de50\",\n    \"VMC:GS_w0WZEvgJF0zf_P4yyTzjjv9oW1z61HHP\",\n
+        \   \"sha512t24u:w0WZEvgJF0zf_P4yyTzjjv9oW1z61HHP\",\n    \"ga4gh:SQ.w0WZEvgJF0zf_P4yyTzjjv9oW1z61HHP\"\n
+        \ ],\n  \"alphabet\": \"ACGNRSTWY\",\n  \"length\": 156040895\n}\n"
+    headers:
+      Connection:
+      - close
+      Content-Length:
+      - '978'
+      Content-Type:
+      - application/json
+      Date:
+      - Fri, 05 Apr 2024 19:12:51 GMT
+      Server:
+      - Werkzeug/2.2.2 Python/3.10.4
+    status:
+      code: 200
+      message: OK
+version: 1

tests/extras/test_cnv_translator.py

@@ -65,7 +65,91 @@ def tlr(rest_dataproxy):
                    'start': 32344742,
                    'type': 'SequenceLocation'},
       'type': 'CopyNumberChange'}
-     )
+    ),
+    (
+        "NC_000023.11:g.(31060227_31100351)_(33274278_33417151)dup",
+        None,
+        {
+            "copyChange": "efo:0030070",
+            "digest": "H0-_q06in6rsvLfq_5b-CSmP4ZQ6r7-Q",
+            "id": "ga4gh:CX.H0-_q06in6rsvLfq_5b-CSmP4ZQ6r7-Q",
+            "location": {
+                "digest": "R3FeXqOiAu8Vms7QngINQwIxW904fdWY",
+                "end": [33274278, 33417151],
+                "id": "ga4gh:SL.R3FeXqOiAu8Vms7QngINQwIxW904fdWY",
+                "sequenceReference": {
+                    "refgetAccession": "SQ.w0WZEvgJF0zf_P4yyTzjjv9oW1z61HHP",
+                    "type": "SequenceReference"
+                },
+                "start": [31060226, 31100350],
+                "type": "SequenceLocation"
+            },
+            "type": "CopyNumberChange"
+        }
+    ),
+    (
+        "NC_000009.11:g.(?_108337304)_(108337428_?)del",
+        None,
+        {
+            "copyChange": "efo:0030067",
+            "digest": "ANthOqEGX8MIn0kXuyQcn9bouYfbFgjH",
+            "id": "ga4gh:CX.ANthOqEGX8MIn0kXuyQcn9bouYfbFgjH",
+            "location": {
+                "digest": "lpDGeQvnz80iis8xSxoCX_Pulnu7wx4M",
+                "end": [108337428, None],
+                "id": "ga4gh:SL.lpDGeQvnz80iis8xSxoCX_Pulnu7wx4M",
+                "sequenceReference": {
+                    "refgetAccession": "SQ.HBckYGQ4wYG9APHLpjoQ9UUe9v7NxExt",
+                    "type": "SequenceReference"
+                },
+                "start": [None, 108337303],
+                "type": "SequenceLocation"
+            },
+            "type": "CopyNumberChange"
+        }
+    ),
+    (
+        "NC_000005.9:g.(90136803)_(90159675)dup",
+        None,
+        {
+            "copyChange": "efo:0030070",
+            "digest": "YcbXUe21Bt1wQDV7zGM0lacOupkxduFS",
+            "id": "ga4gh:CX.YcbXUe21Bt1wQDV7zGM0lacOupkxduFS",
+            "location": {
+                "digest": "r82CARuf8IxOidMdvQCUcsXNp3XiHEVH",
+                "end": 90159675,
+                "id": "ga4gh:SL.r82CARuf8IxOidMdvQCUcsXNp3XiHEVH",
+                "sequenceReference": {
+                    "refgetAccession": "SQ.vbjOdMfHJvTjK_nqvFvpaSKhZillW0SX",
+                    "type": "SequenceReference"
+                },
+                "start": 90136802,
+                "type": "SequenceLocation"
+            },
+            "type": "CopyNumberChange"
+        }
+    ),
+    (
+        "NC_000009.11:g.108337304_(108337428_?)del",
+        None,
+        {
+            "copyChange": "efo:0030067",
+            "digest": "brfJaiKCnSw-mvc3K9sUIEAyCN620PuD",
+            "id": "ga4gh:CX.brfJaiKCnSw-mvc3K9sUIEAyCN620PuD",
+            "location": {
+                "digest": "6myLdODZ8WgbEDXc3HLp88ZbG536NCM-",
+                "end": [108337428, None],
+                "id": "ga4gh:SL.6myLdODZ8WgbEDXc3HLp88ZbG536NCM-",
+                "sequenceReference": {
+                    "refgetAccession": "SQ.HBckYGQ4wYG9APHLpjoQ9UUe9v7NxExt",
+                    "type": "SequenceReference"
+                },
+                "start": 108337303,
+                "type": "SequenceLocation"
+            },
+            "type": "CopyNumberChange"
+        }
+    )
 )
 
 
@@ -76,6 +160,15 @@ def test_from_hgvs_cx(tlr, hgvsexpr ,copy_change, expected):
     cx = tlr._from_hgvs(hgvsexpr, copy_change=copy_change)
     assert cx.model_dump(exclude_none=True) == expected
 
+@pytest.mark.vcf
+def test_from_hgvs_cx_invalid(tlr):
+    """test that _from_hgvs works correctly for copy number change invalid input"""
+    # Should fail since it's not g. or m.
+    with pytest.raises(
+        ValueError, match="Only 'g' and 'm' reference sequences are supported"
+    ):
+        tlr._from_hgvs("NM_001197320.1:c.281_283dup")
+
 
 from_hgvs_cn_tests = (
     ("NC_000013.11:g.26440969_26443305del", 1,