#1217 - T2T - Classification grid dies due to no cached column - don'…

…t try and look for it. Also fixed genome patch lookup

classification/views/classification_datatables.py

@@ -1,3 +1,4 @@
+import logging
 import operator
 from functools import cached_property, reduce
 from typing import Dict, Any, List, Optional
@@ -58,11 +59,14 @@ def render_c_hgvs(self, row: Dict[str, Any]) -> JsonDataType:
         def get_preferred_chgvs_json() -> Dict:
             nonlocal row
             for index, genome_build in enumerate(self.genome_build_prefs):
-                if c_hgvs_string := row.get(ClassificationModification.column_name_for_build(genome_build)):
-                    c_hgvs = CHGVS(c_hgvs_string)
-                    c_hgvs.genome_build = genome_build
-                    c_hgvs.is_desired_build = index == 0
-                    return c_hgvs.to_json()
+                try:
+                    if c_hgvs_string := row.get(ClassificationModification.column_name_for_build(genome_build)):
+                        c_hgvs = CHGVS(c_hgvs_string)
+                        c_hgvs.genome_build = genome_build
+                        c_hgvs.is_desired_build = index == 0
+                        return c_hgvs.to_json()
+                except ValueError:
+                    pass
 
             c_hgvs = CHGVS(row.get('published_evidence__c_hgvs__value'))
             c_hgvs.is_normalised = False
@@ -133,7 +137,18 @@ def __init__(self, request: HttpRequest):
         super().__init__(request)
 
         user_settings = UserSettings.get_for_user(self.user)
-        genome_build_preferred = self.genome_build_prefs[0]
+        # Only GRCh37 and GRCh38 are currently supported. If user has T2T we are going to fall back
+        c_hgvs = None
+        genomic_sort = None
+        for genome_build in self.genome_build_prefs:
+            try:
+                c_hgvs = ClassificationModification.column_name_for_build(genome_build)
+                genomic_sort = ClassificationModification.column_name_for_build(genome_build, 'genomic_sort')
+            except ValueError:
+                pass
+        if not (c_hgvs and genomic_sort):
+            possible_builds = ', '.join((str(gb) for gb in self.genome_build_prefs))
+            raise ValueError(f"Couldn't find c_hgvs and genomic_sort in builds: {possible_builds}")
 
         self.rich_columns = [
             RichColumn(
@@ -164,9 +179,8 @@ def __init__(self, request: HttpRequest):
                 orderable=True
             ),
             RichColumn(
-                key=ClassificationModification.column_name_for_build(genome_build_preferred),
-                # sort_keys=['variant_sort', 'c_hgvs'],  # annotated column
-                sort_keys=[ClassificationModification.column_name_for_build(genome_build_preferred, 'genomic_sort'), 'c_hgvs'],
+                key=c_hgvs,
+                sort_keys=[genomic_sort, 'c_hgvs'],
                 name='c_hgvs',
                 label=f'HGVS ({user_settings.default_genome_build.name})',
                 renderer=self.render_c_hgvs,
@@ -284,8 +298,11 @@ def get_initial_queryset(self) -> QuerySet[ClassificationModification]:
         # user's normalised preference (e.g. 37), the alternative normalised (e.g. 38) the imported c.hgvs
         whens = []
         for genome_build in self.genome_build_prefs:
-            column_name = ClassificationModification.column_name_for_build(genome_build)
-            whens.append(When(**{f'{column_name}__isnull': False, 'then': column_name}))
+            try:
+                column_name = ClassificationModification.column_name_for_build(genome_build)
+                whens.append(When(**{f'{column_name}__isnull': False, 'then': column_name}))
+            except ValueError as ve:
+                logging.warning(ve)
         case = Case(*whens, default=KeyTextTransform('value', KeyTransform('c_hgvs', 'published_evidence')),
                     output_field=TextField())
         initial_qs = initial_qs.annotate(c_hgvs=case)

snpdb/models/models_genome.py

@@ -301,7 +301,12 @@ class GenomeBuildPatchVersion(models.Model):
     patch_version = models.IntegerField(blank=True, null=True)
     """ The version of the patch, or None if the version is unknown, note this is different to an explicit patch_version of 0 """
 
-    GENOME_BUILD_VERSION_RE = re.compile(r"(?P<genome_build>[A-Za-z0-9]+)(?:[.]p(?P<patch_version>[0-9]+))?", re.IGNORECASE)
+    GENOME_BUILD_VERSION_REGEXES = [
+        # Standard eg GRCh37.p13 or GRCh38.p14
+        re.compile(r"(?P<genome_build>[A-Za-z0-9]+)(?:[.]p(?P<patch_version>[0-9]+))?", re.IGNORECASE),
+        # T2T-CHM13v2.0
+        re.compile(r"(?P<genome_build>[A-Za-z0-9-]+)(v(?P<patch_version>[0-9]+\.[0-9]+))", re.IGNORECASE),
+    ]
 
     class Meta:
         unique_together = ('genome_build', 'patch_version')
@@ -316,24 +321,25 @@ def sort_key(obj: GenomeBuildPatchVersion):
 
     @staticmethod
     def get_or_create(name: str) -> 'GenomeBuildPatchVersion':
-        if match := GenomeBuildPatchVersion.GENOME_BUILD_VERSION_RE.match(name):
-            genome_build = GenomeBuild.get_name_or_alias(match.group('genome_build'))
-            patch_version: Optional[int] = None
-            normalized_name: str
-            if patch_version_str := match.group('patch_version'):
-                patch_version = int(patch_version_str)
-                normalized_name = f"{genome_build.name}.p{patch_version}"
-            else:
-                normalized_name = genome_build.name
-
-            gbpv, _ = GenomeBuildPatchVersion.objects.get_or_create(
-                name=normalized_name,
-                genome_build=genome_build,
-                patch_version=patch_version
-            )
-            return gbpv
-        else:
-            raise ValueError(f"Invalid Genome Build Patch Version \"{name}\"")
+        for regex in GenomeBuildPatchVersion.GENOME_BUILD_VERSION_REGEXES:
+            if match := regex.match(name):
+                genome_build = GenomeBuild.get_name_or_alias(match.group('genome_build'))
+                patch_version: Optional[int] = None
+                normalized_name: str
+                if patch_version_str := match.group('patch_version'):
+                    patch_version = int(patch_version_str)
+                    normalized_name = f"{genome_build.name}.p{patch_version}"
+                else:
+                    normalized_name = genome_build.name
+
+                gbpv, _ = GenomeBuildPatchVersion.objects.get_or_create(
+                    name=normalized_name,
+                    genome_build=genome_build,
+                    patch_version=patch_version
+                )
+                return gbpv
+
+        raise ValueError(f"Invalid Genome Build Patch Version '{name}'")
 
     @staticmethod
     def get_unspecified_patch_version_for(genome_build: GenomeBuild) -> 'GenomeBuildPatchVersion':