Add support for mito m. in translate_from (#362)

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

@@ -46,10 +46,9 @@ class Translator:
         r"(?P<chr>[^-]+)-(?P<pos>\d+)-(?P<ref>[ACGTURYKMSWBDHVN]+)-(?P<alt>[ACGTURYKMSWBDHVN]+)",
         re.IGNORECASE
     )
-    hgvs_re = re.compile(r"[^:]+:[cgnpr]\.")
+    hgvs_re = re.compile(r"[^:]+:[cgmnpr]\.")
     spdi_re = re.compile(r"(?P<ac>[^:]+):(?P<pos>\d+):(?P<del_len_or_seq>\w*):(?P<ins_seq>\w*)")
 
-
     def __init__(
         self,
         data_proxy,
@@ -86,17 +85,34 @@ def _get_hgvs_refget_ac(self, sv: hgvs.sequencevariant.SequenceVariant):
 
     @staticmethod
     def _ir_stype(a):
-        """Get accession's sequence type"""
-        if a.startswith("refseq:NM_"):
-            return "n"
-        if a.startswith("refseq:NP_"):
-            return "p"
-        if a.startswith("refseq:NG_"):
-            return "g"
-        if a.startswith("refseq:NC_"):
-            return "g"
-        if a.startswith("GRCh"):
-            return "g"
+        """
+        The purpose of this function is to provide a convenient way to extract the sequence type from an accession by matching its prefix to a known set of prefixes.
+
+        Args:
+        a (str): The accession string.
+
+        Returns:
+        str or None: The sequence type associated with the accession string, or None if no matching prefix is found.
+        """
+
+        prefix_dict = {
+            "refseq:NM_": "n",
+            "refseq:NC_012920": "m",
+            "refseq:NG_": "g",
+            "refseq:NC_00": "g",
+            "refseq:NW_": "g",
+            "refseq:NT_": "g",
+            "refseq:NR_": "n",
+            "refseq:NP_": "p",
+            "refseq:XM_": "n",
+            "refseq:XR_": "n",
+            "refseq:XP_": "p",
+            "GRCh": "g",
+        }
+
+        for prefix, stype in prefix_dict.items():
+            if a.startswith(prefix):
+                return stype
         return None
 
     def translate_from(self, var, fmt=None, **kwargs):
@@ -152,7 +168,6 @@ def translate_to(self, vo, fmt):
         t = self.to_translators[fmt]
         return t(vo)
 
-
     ############################################################################
     # INTERNAL
 
@@ -580,7 +595,7 @@ def _to_hgvs(self, vo, namespace="refseq"):
             if ns.startswith("GRC") and namespace is None:
                 continue
 
-            if not (any(a.startswith(pfx) for pfx in ("NM", "NP", "NC", "NG"))):
+            if not (any(a.startswith(pfx) for pfx in ("NM", "NP", "NC", "NG", "NR", "NW", "NT", "XM", "XR", "XP"))):
                 continue
 
             var.ac = a
@@ -727,7 +742,7 @@ def _post_process_imported_cnv(self, copy_number):
 
     from ga4gh.vrs.dataproxy import create_dataproxy
     # dp = create_dataproxy("seqrepo+file:///usr/local/share/seqrepo/latest")
-    dp = create_dataproxy("seqrepo + http://localhost:5555/seqrepo")
+    dp = create_dataproxy("seqrepo + http://localhost:5000/seqrepo")
     tlr = Translator(data_proxy=dp)
 
     expressions = [

tests/extras/cassettes/test_from_beacon.yaml

@@ -9,7 +9,7 @@ interactions:
       Connection:
       - keep-alive
       User-Agent:
-      - python-requests/2.28.2
+      - python-requests/2.31.0
     method: GET
     uri: http://localhost:5000/seqrepo/1/metadata/GRCh38:19
   response:
@@ -35,7 +35,54 @@ interactions:
       Content-Type:
       - application/json
       Date:
-      - Mon, 19 Feb 2024 07:31:12 GMT
+      - Wed, 13 Mar 2024 11:29:22 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:MT
+  response:
+    body:
+      string: "{\n  \"added\": \"2016-08-24T06:13:07Z\",\n  \"aliases\": [\n    \"Ensembl:MT\",\n
+        \   \"ensembl:MT\",\n    \"GRCh37.p10:MT\",\n    \"GRCh37.p10:chrM\",\n    \"GRCh37.p11:MT\",\n
+        \   \"GRCh37.p11:chrM\",\n    \"GRCh37.p12:MT\",\n    \"GRCh37.p12:chrM\",\n
+        \   \"GRCh37.p13:MT\",\n    \"GRCh37.p13:chrM\",\n    \"GRCh37.p2:MT\",\n
+        \   \"GRCh37.p2:chrM\",\n    \"GRCh37.p5:MT\",\n    \"GRCh37.p5:chrM\",\n
+        \   \"GRCh37.p9:MT\",\n    \"GRCh37.p9:chrM\",\n    \"GRCh38:MT\",\n    \"GRCh38:chrM\",\n
+        \   \"GRCh38.p1:MT\",\n    \"GRCh38.p1:chrM\",\n    \"GRCh38.p10:MT\",\n    \"GRCh38.p10:chrM\",\n
+        \   \"GRCh38.p11:MT\",\n    \"GRCh38.p11:chrM\",\n    \"GRCh38.p12:MT\",\n
+        \   \"GRCh38.p12:chrM\",\n    \"GRCh38.p2:MT\",\n    \"GRCh38.p2:chrM\",\n
+        \   \"GRCh38.p3:MT\",\n    \"GRCh38.p3:chrM\",\n    \"GRCh38.p4:MT\",\n    \"GRCh38.p4:chrM\",\n
+        \   \"GRCh38.p5:MT\",\n    \"GRCh38.p5:chrM\",\n    \"GRCh38.p6:MT\",\n    \"GRCh38.p6:chrM\",\n
+        \   \"GRCh38.p7:MT\",\n    \"GRCh38.p7:chrM\",\n    \"GRCh38.p8:MT\",\n    \"GRCh38.p8:chrM\",\n
+        \   \"GRCh38.p9:MT\",\n    \"GRCh38.p9:chrM\",\n    \"MD5:c68f52674c9fb33aef52dcf399755519\",\n
+        \   \"NCBI:NC_012920.1\",\n    \"refseq:NC_012920.1\",\n    \"SEGUID:eQNFYXnsCzhp/MkfBUBVnuFZzTA\",\n
+        \   \"SHA1:7903456179ec0b3869fcc91f0540559ee159cd30\",\n    \"VMC:GS_k3grVkjY-hoWcCUojHw6VU6GE3MZ8Sct\",\n
+        \   \"sha512t24u:k3grVkjY-hoWcCUojHw6VU6GE3MZ8Sct\",\n    \"ga4gh:SQ.k3grVkjY-hoWcCUojHw6VU6GE3MZ8Sct\",\n
+        \   \"hs37-1kg:MT\",\n    \"hs37d5:MT\"\n  ],\n  \"alphabet\": \"ACGNT\",\n
+        \ \"length\": 16569\n}\n"
+    headers:
+      Connection:
+      - close
+      Content-Length:
+      - '1355'
+      Content-Type:
+      - application/json
+      Date:
+      - Wed, 13 Mar 2024 11:29:22 GMT
       Server:
       - Werkzeug/2.2.2 Python/3.10.4
     status: