Merge pull request #855 from SACGF/#3485_stop_MedGen_termm_trigger_hgvs

#3485 handled search error for MedGen and providing external Link.

ontology/models/models_ontology.py

@@ -46,14 +46,16 @@ class OntologyService(models.TextChoices):
 
     DOID = "DOID", "DOID"
     ORPHANET = "Orphanet", "Orphanet"
+    MEDGEN = "MedGen", "MedGen"
 
     EXPECTED_LENGTHS: Dict[str, int] = Constant({
         MONDO[0]: 7,
         OMIM[0]: 6,
         HPO[0]: 7,
         HGNC[0]: 1,  # HGNC ids aren't typically 0 padded, because they're not monsters
         DOID[0]: None,  # variable length with padded 0s
-        ORPHANET[0]: 1  # ORPHANET ids aren't typically 0 padded
+        ORPHANET[0]: 1,  # ORPHANET ids aren't typically 0 padded
+        MEDGEN[0]: None
     })
 
     IMPORTANCE: Dict[str, int] = Constant({
@@ -62,7 +64,8 @@ class OntologyService(models.TextChoices):
         HPO[0]: 4,  # put HPO relationships last as they occasionally spam OMIM
         DOID[0]: 5,
         ORPHANET[0]: 6,
-        HGNC[0]: 1  # show gene relationships first
+        HGNC[0]: 1,  # show gene relationships first
+        MEDGEN[0]: 7
     })
 
     URLS: Dict[str, str] = Constant({
@@ -71,7 +74,8 @@ class OntologyService(models.TextChoices):
         HPO[0]: "https://hpo.jax.org/app/browse/term/HP:${1}",
         HGNC[0]: "https://www.genenames.org/data/gene-symbol-report/#!/hgnc_id/HGNC:${1}",
         DOID[0]: "https://www.ebi.ac.uk/ols/ontologies/doid/terms?iri=http%3A%2F%2Fpurl.obolibrary.org%2Fobo%2FDOID_${1}",
-        ORPHANET[0]: "https://www.orpha.net/consor/cgi-bin/OC_Exp.php?lng=EN&Expert=${1}"
+        ORPHANET[0]: "https://www.orpha.net/consor/cgi-bin/OC_Exp.php?lng=EN&Expert=${1}",
+        MEDGEN[0]: "https://www.ncbi.nlm.nih.gov/medgen/${1}"
     })
 
     LOCAL_ONTOLOGY_PREFIXES: Set[str] = Constant({
@@ -86,7 +90,8 @@ class OntologyService(models.TextChoices):
         OMIM[0],
         HPO[0],
         DOID[0],
-        ORPHANET[0]
+        ORPHANET[0],
+        MEDGEN[0]
     })
 
     @staticmethod
@@ -239,6 +244,12 @@ class OntologyIdNormalized:
     def num_part(self) -> int:
         return int(self.postfix)
 
+    def num_part_safe(self) -> int:
+        try:
+            return self.num_part
+        except:
+            return 0
+
     @staticmethod
     def normalize(dirty_id: str) -> 'OntologyIdNormalized':
         parts = re.split("[:|_]", dirty_id)
@@ -250,17 +261,18 @@ def normalize(dirty_id: str) -> 'OntologyIdNormalized':
             prefix = "Orphanet"
         elif prefix.upper() == "MIM":
             prefix = "OMIM"
+        elif prefix.upper() == "MEDGEN":
+            prefix = "MedGen"
         prefix = OntologyService(prefix)
         postfix = parts[1].strip()
         try:
-            num_part = int(postfix)
+            num_part = str(postfix)
             clean_id: str
             if OntologyService.EXPECTED_LENGTHS[prefix]:
                 clean_id = OntologyService.index_to_id(prefix, num_part)
             else:
                 # variable length IDs like DOID
                 clean_id = f"{prefix}:{postfix}"
-
             return OntologyIdNormalized(prefix=prefix, postfix=postfix, full_id=clean_id, clean=True)
 
         except ValueError:
@@ -443,10 +455,14 @@ def get_or_stub(id_str: Union[str, OntologyIdNormalized]) -> 'OntologyTerm':
         if normal_id.clean:
             if existing := OntologyTerm.objects.filter(id=normal_id.full_id).first():
                 return existing
+            try:
+                index_num_part_value = normal_id.num_part
+            except :
+                index_num_part_value = normal_id.num_part_safe  # Ontologies like MedGen can have alpha characters in the "index", providing an index of 0 until we update the model
             return OntologyTerm(
                 id=normal_id.full_id,
                 ontology_service=normal_id.prefix,
-                index=normal_id.num_part,
+                index=index_num_part_value,
                 name=""
             )
         else:

ontology/models/ontology_search.py

@@ -27,8 +27,7 @@ def validate_ontology(term: OntologyTerm, preview_proxy: Optional[PreviewProxyMo
     return SearchResult(preview, messages=messages)
 
 
-ONTOLOGY_TERM_PATTERN = re.compile(r"^(MONDO|OMIM|MIM|HPO|HP|DOID|ORPHANET)\s*:\s*[0-9]+$", re.IGNORECASE)
-
+ONTOLOGY_TERM_PATTERN = re.compile(r"^((MONDO|OMIM|MIM|HPO|HP|DOID|ORPHANET)\s*:\s*[0-9]+)|(MEDGEN\s*:\s*\w{0,2}[0-9]+)$", re.IGNORECASE)
 
 @search_receiver(
     search_type=OntologyTerm,

snpdb/search.py

@@ -794,6 +794,11 @@ def search_data(user: User, search_string: str, classify: bool = False) -> Searc
     return SearchResponsesCombined(search_input, responses)
 
 
+INVALID_INPUT = object()
+"""
+Return this (and only this) from a search if you want to act as if the search pattern wasn't met 
+"""
+
 def search_receiver(
         search_type: Optional[PreviewCoordinator],
         pattern: Pattern = HAS_ANYTHING,
@@ -870,7 +875,10 @@ def search_func(sender: Any, search_input: SearchInput, **kwargs):
                     # as Variants get merged into Alleles, we want to avoid limiting them (except under extreme conditions)
                     limit = MAX_VARIANT_RESULTS if search_type.preview_category() == "Variant" else MAX_RESULTS_PER_SEARCH
                     for result in func(SearchInputInstance(expected_type=search_type, search_input=search_input, match=match)):
-                        if result is None:
+                        if result == INVALID_INPUT:
+                            matched_pattern = False
+                            break
+                        elif result is None:
                             raise ValueError(f"Search {sender.__name__} returned None")
                         elif isinstance(result, SearchMessageOverall):
                             overall_messages.add(result)

snpdb/signals/variant_search.py

@@ -23,7 +23,7 @@
 from snpdb.models import Variant, LOCUS_PATTERN, LOCUS_NO_REF_PATTERN, DbSNP, DBSNP_PATTERN, VariantCoordinate, \
     ClinGenAllele, GenomeBuild, Contig, HGVS_UNCLEANED_PATTERN
 from snpdb.search import search_receiver, SearchInputInstance, SearchExample, SearchResult, SearchMessageOverall, \
-    SearchMessage
+    SearchMessage, INVALID_INPUT
 from upload.models import ModifiedImportedVariant
 
 COSMIC_PATTERN = re.compile(r"^(COS[VM])[0-9]{3,}$", re.IGNORECASE)
@@ -398,7 +398,6 @@ def _search_hgvs_using_gene_symbol(
         elif not (settings.SEARCH_HGVS_GENE_SYMBOL_USE_MANE or settings.SEARCH_HGVS_GENE_SYMBOL_USE_ALL_TRANSCRIPTS):
             yield SearchMessageOverall("\n".join(messages_as_strs))
 
-
 @search_receiver(
     search_type=Variant,
     pattern=HGVS_UNCLEANED_PATTERN,
@@ -409,6 +408,8 @@ def _search_hgvs_using_gene_symbol(
     )
 )
 def search_hgvs(search_input: SearchInputInstance) -> Iterable[SearchResult]:
+    if search_input.search_string.lower().startswith("medgen"):
+        return [INVALID_INPUT]
     for_all_genome_builds = []
     for genome_build in search_input.genome_builds:
         for_all_genome_builds.append(_search_hgvs(hgvs_string=search_input.search_string, user=search_input.user, genome_build=genome_build, visible_variants=search_input.get_visible_variants(genome_build), classify=search_input.classify))