Merge pull request #1277 from mild-blue/abragtim/fix-frequency-for-un…

…known-hlas Fix frequency for unknown hlas
mild-blue · Aug 14, 2024 · b06b604 · b06b604
2 parents 3f2ca22 + 3ed8113
commit b06b604
Show file tree

Hide file tree

Showing 4 changed files with 41 additions and 4 deletions.
diff --git a/tests/web/test_do_crossmatch_api.py b/tests/web/test_do_crossmatch_api.py
@@ -230,6 +230,38 @@ def test_do_crossmatch_api_with_different_code_formats(self):
             self.assertCountEqual(expected_assumed_hla_types,
                                   res.json['hla_to_antibody'][0]['assumed_hla_types'])
 
+        # CASE: donor has unknown HIGH RES code according to our data
+        json = {
+            'potential_donor_hla_typing': [[{'hla_code': 'C*07:9999', 'is_frequent': False}]],
+            'recipient_antibodies': [],
+        }
+        with self.app.test_client() as client:
+            res = client.post(f'{API_VERSION}/{CROSSMATCH_NAMESPACE}/do-crossmatch', json=json,
+                              headers=self.auth_headers)
+            # since we don't know the SPLIT for this code, we'll leave HIGH RES,
+            # but designate its frequency as the input
+            self.assertEqual(200, res.status_code)
+            expected_assumed_hla_types = [asdict(create_hla_type_with_frequency(
+                HLATypeWithFrequencyRaw('C*07:9999', is_frequent=False)
+            ))]
+            self.assertCountEqual(expected_assumed_hla_types,
+                                  res.json['hla_to_antibody'][0]['assumed_hla_types'])
+
+        # CASE: donor has unknown code in unknown format according to our data
+        json = {
+            'potential_donor_hla_typing': [[{'hla_code': '921HLAUNKNWONFORMAT123', 'is_frequent': False}]],
+            'recipient_antibodies': [],
+        }
+        with self.app.test_client() as client:
+            res = client.post(f'{API_VERSION}/{CROSSMATCH_NAMESPACE}/do-crossmatch', json=json,
+                              headers=self.auth_headers)
+            self.assertEqual(200, res.status_code)
+            expected_assumed_hla_types = [asdict(create_hla_type_with_frequency(
+                HLATypeWithFrequencyRaw('921HLAUNKNWONFORMAT123', is_frequent=False)
+            ))]
+            self.assertCountEqual(expected_assumed_hla_types,
+                                  res.json['hla_to_antibody'][0]['assumed_hla_types'])
+
     def test_theoretical_and_double_antibodies(self):
         json = {
             'potential_donor_hla_typing': [[{'hla_code': 'DPA1*01:03', 'is_frequent': True}],
@@ -967,7 +999,7 @@ def test_summary(self):
                                       'cutoff': 2000
                                       },
                                      {'mfi': 2000,
-                                      'name': 'DPA1*01:02',
+                                      'name': 'DPA1*01:07',
                                       'cutoff': 2000}] + type_A_antibodies
         }
 

diff --git a/txmatching/utils/hla_system/hla_regexes.py b/txmatching/utils/hla_system/hla_regexes.py
@@ -1,7 +1,7 @@
 import re
 from typing import Optional
 
-HIGH_RES_REGEX_STRING = r'^([A-Z]+\d?\*\d{2,4}(:\d{2,3}))(:\d{2,3})*'
+HIGH_RES_REGEX_STRING = r'^([A-Z]+\d?\*\d{2,4}(:\d{2,4}))(:\d{2,4})*'
 HIGH_RES_REGEX = re.compile(HIGH_RES_REGEX_STRING+r'$')  # with high res subunit
 HIGH_RES_REGEX_ENDING_WITH_LETTER = re.compile(HIGH_RES_REGEX_STRING+r'([A-Z])$')
 HIGH_RES_WITH_SUBUNITS_REGEX = re.compile(r'(DP|DQ)\d?\[(\d{2}:\d{2}),(\d{2}:\d{2})]')

diff --git a/txmatching/utils/hla_system/hla_transformations/hla_transformations.py b/txmatching/utils/hla_system/hla_transformations/hla_transformations.py
@@ -75,9 +75,9 @@ def _process_hla_in_high_res(hla_high_res: str, hla_raw_code: str) -> HlaCodePro
     )
     if exception_split_broad_code is None:
         if hla_raw_code in ALL_ULTRA_HIGH_RES_CODES:
-            return process_parsing_result(hla_raw_code, hla_raw_code, ParsingIssueDetail.UNKNOWN_TRANSFORMATION_FROM_HIGH_RES)
+            return process_parsing_result(hla_raw_code, None, ParsingIssueDetail.UNKNOWN_TRANSFORMATION_FROM_HIGH_RES)
         else:
-            return process_parsing_result(hla_raw_code, hla_raw_code, ParsingIssueDetail.UNPARSABLE_HLA_CODE)
+            return process_parsing_result(hla_raw_code, None, ParsingIssueDetail.UNPARSABLE_HLA_CODE)
     if isinstance(exception_split_broad_code, ParsingIssueDetail):
         return process_parsing_result(hla_raw_code, hla_raw_code, exception_split_broad_code)
     if hla_high_res in ALL_HIGH_RES_CODES_WITH_ASSUMED_SPLIT_BROAD_CODE:

diff --git a/txmatching/web/api/crossmatch_api.py b/txmatching/web/api/crossmatch_api.py
@@ -260,6 +260,11 @@ def _convert_potential_hla_types_to_low_res(
     assumed_hla_types_raw = {HLATypeWithFrequencyRaw(
         hla_code=potential_hla_type.hla_type.code.get_low_res_code(),
         is_frequent=True
+        # if the code does not exist in our database and has an unknown format,
+        # we cannot determine whether it is a high or low res code,
+        # then leave the is_frequent as it is in the input
+        if potential_hla_type.hla_type.code.high_res != potential_hla_type.hla_type.code.get_low_res_code()
+        else potential_hla_type.is_frequent
     ) for potential_hla_type in potential_hla_types}
     assumed_hla_types = [create_hla_type_with_frequency(assumed_hla_type_raw)
                          for assumed_hla_type_raw in assumed_hla_types_raw