build!: update vicc normalizers to pull in gks coding changes (#432)

close #423
cancervariants · Mar 9, 2025 · 3a1f3ea · 3a1f3ea
1 parent 6550d7c
commit 3a1f3ea
Show file tree

Hide file tree

Showing 9 changed files with 227 additions and 155 deletions.
diff --git a/Pipfile b/Pipfile
@@ -4,11 +4,11 @@ verify_ssl = true
 name = "pypi"
 
 [packages]
-"ga4gh.vrs" = "~=2.0.0a10"
-gene-normalizer = {version = "~=0.4.1", extras = ["etl"]}
-variation-normalizer = "~=0.10.0"
-disease-normalizer = {version = "~=0.5.0", extras = ["etl"]}
-thera-py = {version = "~=0.6.0", extras = ["etl"]}
+"ga4gh.vrs" = "==2.0.0a13"
+gene-normalizer = {version = "~=0.7.0", extras = ["etl"]}
+variation-normalizer = "~=0.12.1"
+disease-normalizer = {version = "~=0.8.0", extras = ["etl"]}
+thera-py = {version = "~=0.8.0", extras = ["etl"]}
 civicpy = "~=3.1"
 requests = "*"
 pydantic = "==2.*"

diff --git a/pyproject.toml b/pyproject.toml
@@ -26,10 +26,10 @@ dependencies = [
     "ga4gh.vrs==2.0.0a13",
     "ga4gh.cat_vrs~=0.2.1",
     "ga4gh.va_spec~=0.2.0a0",
-    "gene-normalizer[etl]~=0.6.0",
-    "variation-normalizer~=0.12.0",
-    "disease-normalizer[etl]~=0.7.0",
-    "thera-py[etl]~=0.7.1",
+    "gene-normalizer[etl]~=0.7.0",
+    "variation-normalizer~=0.12.1",
+    "disease-normalizer[etl]~=0.8.0",
+    "thera-py[etl]~=0.8.0",
     "civicpy~=3.1",
     "requests",
     "pydantic==2.*",

diff --git a/src/metakb/load_data.py b/src/metakb/load_data.py
@@ -47,7 +47,10 @@ def _add_mappings_and_exts_to_obj(obj: dict, obj_keys: list[str]) -> None:
                 extensions = mapping.get("extensions") or []
                 for ext in extensions:
                     if ext["name"] == NormalizerExtensionName.PRIORITY and ext["value"]:
-                        normalizer_id = mapping["coding"]["code"]
+                        if mapping["coding"]["id"].startswith("MONDO"):
+                            normalizer_id = mapping["coding"]["code"]
+                        else:
+                            normalizer_id = mapping["coding"]["id"]
                         obj["normalizer_id"] = normalizer_id
                         obj_keys.append("normalizer_id:$normalizer_id")
                         break

diff --git a/src/metakb/transformers/base.py b/src/metakb/transformers/base.py
@@ -563,6 +563,7 @@ def _update_mapping(
             mapping: ConceptMapping,
             normalized_id: str,
             normalizer_label: str,
+            match_on_coding_id: bool = True,
         ) -> Extension:
             """Update ``mapping`` to include extension on whether ``mapping`` contains
             code that matches the merged record's primary identifier.
@@ -572,10 +573,16 @@ def _update_mapping(
                 matches normalized merged identifier.
             :param normalized_id: Concept ID from normalized record
             :param normalizer_label: Label from normalized record
+            :param match_on_coding_id: Whether to match on ``coding.id`` or
+                ``coding.code`` (MONDO is represented differently)
             :return: ConceptMapping with normalizer extension added as well as label (
                 if mapping id matches normalized merged id)
             """
-            is_priority = normalized_id == mapping.coding.code.root
+            is_priority = (
+                normalized_id == mapping.coding.id
+                if match_on_coding_id
+                else normalized_id == mapping.coding.code.root.lower()
+            )
 
             merged_id_ext = Extension(
                 name=NormalizerExtensionName.PRIORITY.value, value=is_priority
@@ -599,25 +606,29 @@ def _update_mapping(
 
         normalizer_mappings = normalizer_resp_obj.mappings or []
         for mapping in normalizer_mappings:
-            if normalized_id == mapping.coding.code.root:
+            if normalized_id == mapping.coding.id:
                 mappings.append(
                     _update_mapping(mapping, normalized_id, normalizer_label)
                 )
             else:
-                mapping_code_lower = mapping.coding.code.root.lower()
                 if (
                     is_disease
-                    and mapping_code_lower.startswith(
+                    and mapping.coding.code.root.lower().startswith(
                         DiseaseNamespacePrefix.MONDO.value
                     )
                 ) or (
                     is_gene
-                    and mapping_code_lower.startswith(
+                    and mapping.coding.id.startswith(
                         (GeneNamespacePrefix.NCBI.value, GeneNamespacePrefix.HGNC.value)
                     )
                 ):
                     mappings.append(
-                        _update_mapping(mapping, normalized_id, normalizer_label)
+                        _update_mapping(
+                            mapping,
+                            normalized_id,
+                            normalizer_label,
+                            match_on_coding_id=is_gene,
+                        )
                     )
         return mappings
 

diff --git a/src/metakb/transformers/civic.py b/src/metakb/transformers/civic.py
@@ -858,8 +858,8 @@ def _get_ncbi_concept_mapping(ncbigene_id: str, gene: dict) -> ConceptMapping:
 
                 civic_ncbi_annotation_match = False
                 for mapping in mappings:
-                    if mapping.coding.code.root.startswith("ncbigene:"):
-                        if mapping.coding.code.root == ncbigene:
+                    if mapping.coding.id.startswith("ncbigene:"):
+                        if mapping.coding.id == ncbigene:
                             mapping.extensions.append(
                                 Extension(name="civic_annotation", value=True)
                             )
@@ -869,7 +869,7 @@ def _get_ncbi_concept_mapping(ncbigene_id: str, gene: dict) -> ConceptMapping:
                         _logger.debug(
                             "CIViC NCBI gene and Gene Normalizer mismatch: %s vs %s",
                             ncbigene,
-                            mapping.coding.code.root,
+                            mapping.coding.id,
                         )
 
                 if not civic_ncbi_annotation_match: