feat!: update models to vrs 2.0.0 community review ballot (#582)

close #581 

* Update modules to vrs
[2.0.0-ballot.2024-11.3](https://github.com/ga4gh/vrs/tree/2.0.0-ballot.2024-11.3)
tag
  * `CopyNumberChange` has `mappings` for EFO code

pyproject.toml

@@ -28,8 +28,8 @@ dependencies = [
     "fastapi",
     "uvicorn",
     "pydantic ==2.*",
-    "ga4gh.vrs[extras] ~= 2.0.0a10",
-    "gene-normalizer ~=0.4.0",
+    "ga4gh.vrs[extras] ==2.0.0a13",
+    "gene-normalizer ~=0.6.0",
     "boto3",
     "cool-seq-tool ~=0.6.0",
     "bioutils"

src/variation/gnomad_vcf_to_protein_variation.py

@@ -5,7 +5,8 @@
 from cool_seq_tool.handlers import SeqRepoAccess
 from cool_seq_tool.mappers import ManeTranscript
 from cool_seq_tool.schemas import Strand
-from ga4gh.core import domain_models, ga4gh_identify
+from ga4gh.core import ga4gh_identify
+from ga4gh.core.models import MappableConcept
 from ga4gh.vrs import models, normalize
 from gene.query import QueryHandler as GeneQueryHandler
 from gene.schemas import MatchType as GeneMatchType
@@ -413,14 +414,14 @@ def _get_protein_representation(
             self.seqrepo_access, p_ac, variation.location.start, variation.location.end
         )
         if loc_seq:
-            variation.location.sequence = models.SequenceString(root=loc_seq)
+            variation.location.sequence = models.sequenceString(root=loc_seq)
 
         # Add VRS digests for VRS Allele and VRS Sequence Location
         variation.id = ga4gh_identify(variation)
         variation.location.id = ga4gh_identify(variation.location)
         return variation
 
-    def _get_gene_context(self, gene: str) -> domain_models.Gene | None:
+    def _get_gene_context(self, gene: str) -> MappableConcept | None:
         """Get additional gene information from gene-normalizer
 
         :param gene: Gene symbol

src/variation/hgvs_dup_del_mode.py

@@ -2,11 +2,15 @@
 
 from cool_seq_tool.handlers import SeqRepoAccess
 from cool_seq_tool.schemas import ResidueMode
-from ga4gh.core import entity_models, ga4gh_identify
+from ga4gh.core import ga4gh_identify
+from ga4gh.core.models import (
+    Extension,
+)
 from ga4gh.vrs import models, normalize
 
 from variation.schemas.normalize_response_schema import HGVSDupDelModeOption
 from variation.schemas.token_response_schema import AMBIGUOUS_REGIONS, AltType
+from variation.utils import get_copy_change_concept
 
 # Define deletion alt types
 DELS = {AltType.DELETION_AMBIGUOUS, AltType.DELETION}
@@ -49,7 +53,7 @@ def default_mode(
         baseline_copies: int | None = None,
         copy_change: models.CopyChange | None = None,
         alt: str | None = None,
-        extensions: list[entity_models.Extension] | None = None,
+        extensions: list[Extension] | None = None,
     ) -> dict | None:
         """Use default characteristics to return a variation.
         If baseline_copies not provided and endpoints are ambiguous - copy_number_change
@@ -92,7 +96,7 @@ def copy_number_count_mode(
         alt_type: AltType,
         location: dict,
         baseline_copies: int,
-        extensions: list[entity_models.Extension] | None = None,
+        extensions: list[Extension] | None = None,
     ) -> dict:
         """Return a VRS Copy Number Variation.
 
@@ -119,7 +123,7 @@ def copy_number_change_mode(
         alt_type: AltType,
         location: dict,
         copy_change: models.CopyChange | None = None,
-        extensions: list[entity_models.Extension] | None = None,
+        extensions: list[Extension] | None = None,
     ) -> dict:
         """Return copy number change variation
 
@@ -142,7 +146,9 @@ def copy_number_change_mode(
         seq_loc = models.SequenceLocation(**location)
         seq_loc.id = ga4gh_identify(seq_loc)
         cx = models.CopyNumberChange(
-            location=seq_loc, copyChange=copy_change, extensions=extensions
+            location=seq_loc,
+            copyChange=get_copy_change_concept(copy_change),
+            extensions=extensions,
         )
         cx.id = ga4gh_identify(cx)
         return cx.model_dump(exclude_none=True)
@@ -153,7 +159,7 @@ def allele_mode(
         alt_type: AltType,
         vrs_seq_loc_ac: str,
         alt: str,
-        extensions: list[entity_models.Extension] | None = None,
+        extensions: list[Extension] | None = None,
     ) -> dict | None:
         """Return a VRS Allele with a normalized LiteralSequenceExpression or
         ReferenceLengthExpression.
@@ -208,7 +214,7 @@ def interpret_variation(
         baseline_copies: int | None = None,
         copy_change: models.CopyChange | None = None,
         alt: str | None = None,
-        extensions: list[entity_models.Extension] | None = None,
+        extensions: list[Extension] | None = None,
     ) -> dict:
         """Interpret variation using HGVSDupDelMode
 

src/variation/schemas/copy_number_schema.py

@@ -324,8 +324,8 @@ class ParsedToCxVarService(ServiceResponse):
             "example": {
                 "copy_number_change": {
                     "type": "CopyNumberChange",
-                    "id": "ga4gh:CX.5kaJC-7Jj851bfJ6EipsHV413feg1T4T",
-                    "digest": "5kaJC-7Jj851bfJ6EipsHV413feg1T4T",
+                    "id": "ga4gh:CX.XIsVHbhEUbXraIgpgV4ToCa-6oZWMRUD",
+                    "digest": "XIsVHbhEUbXraIgpgV4ToCa-6oZWMRUD",
                     "location": {
                         "type": "SequenceLocation",
                         "id": "ga4gh:SL.Iz_azSFTEulx7tCluLgGhE1n0hTLUocb",
@@ -337,7 +337,18 @@ class ParsedToCxVarService(ServiceResponse):
                         "start": 10000,
                         "end": 1223133,
                     },
-                    "copyChange": "efo:0030069",
+                    "copyChange": {
+                        "primaryCode": "EFO:0030069",
+                        "mappings": [
+                            {
+                                "coding": {
+                                    "system": "https://www.ebi.ac.uk/efo/",
+                                    "code": "EFO:0030069",
+                                },
+                                "relation": "exactMatch",
+                            }
+                        ],
+                    },
                 },
                 "service_meta_": {
                     "name": "variation-normalizer",
@@ -379,8 +390,8 @@ class AmplificationToCxVarService(ServiceResponse):
                 },
                 "amplification_label": "BRAF Amplification",
                 "copy_number_change": {
-                    "id": "ga4gh:CX._UsXDMCLtPwsVKiNByhbwfS569K1wLWW",
-                    "digest": "_UsXDMCLtPwsVKiNByhbwfS569K1wLWW",
+                    "id": "ga4gh:CX.uPQaLz6KSwXWdsjNUZ5kRn3znBZF5YwV",
+                    "digest": "uPQaLz6KSwXWdsjNUZ5kRn3znBZF5YwV",
                     "type": "CopyNumberChange",
                     "location": {
                         "id": "ga4gh:SL.0nPwKHYNnTmJ06G-gSmz8BEhB_NTp-0B",
@@ -393,7 +404,18 @@ class AmplificationToCxVarService(ServiceResponse):
                         "start": 140713327,
                         "end": 140924929,
                     },
-                    "copyChange": "efo:0030072",
+                    "copyChange": {
+                        "primaryCode": "EFO:0030072",
+                        "mappings": [
+                            {
+                                "coding": {
+                                    "system": "https://www.ebi.ac.uk/efo/",
+                                    "code": "EFO:0030072",
+                                },
+                                "relation": "exactMatch",
+                            }
+                        ],
+                    },
                 },
                 "service_meta_": {
                     "version": __version__,

src/variation/schemas/gnomad_vcf_to_protein_schema.py

@@ -1,11 +1,11 @@
 """Module for gnomad vcf to protein response schema"""
 
-from ga4gh.core import domain_models
+from ga4gh.core.models import MappableConcept
 
 from variation.schemas.normalize_response_schema import NormalizeService
 
 
 class GnomadVcfToProteinService(NormalizeService):
     """Define response for gnomad vcf to protein service"""
 
-    gene_context: domain_models.Gene | None = None
+    gene_context: MappableConcept | None = None

src/variation/schemas/hgvs_to_copy_number_schema.py

@@ -56,8 +56,8 @@ class HgvsToCopyNumberChangeService(ServiceResponse):
             "example": {
                 "hgvs_expr": "NC_000003.12:g.49531262dup",
                 "copy_number_change": {
-                    "id": "ga4gh:CX.Zzws_y4cnoooQ7WXjg2B3nKIyFWXzOg3",
-                    "digest": "Zzws_y4cnoooQ7WXjg2B3nKIyFWXzOg3",
+                    "id": "ga4gh:CX.30bDl5yhHzjc4M5uGS_8IeYMzHQksQGh",
+                    "digest": "30bDl5yhHzjc4M5uGS_8IeYMzHQksQGh",
                     "type": "CopyNumberChange",
                     "location": {
                         "id": "ga4gh:SL.2vbgFGHGB0QGODwgZNi05fWbROkkjf04",
@@ -70,7 +70,18 @@ class HgvsToCopyNumberChangeService(ServiceResponse):
                         "start": 49531261,
                         "end": 49531262,
                     },
-                    "copyChange": "efo:0030069",
+                    "copyChange": {
+                        "primaryCode": "EFO:0030069",
+                        "mappings": [
+                            {
+                                "coding": {
+                                    "system": "https://www.ebi.ac.uk/efo/",
+                                    "code": "EFO:0030069",
+                                },
+                                "relation": "exactMatch",
+                            }
+                        ],
+                    },
                 },
                 "service_meta_": {
                     "name": "variation-normalizer",

src/variation/schemas/token_response_schema.py

@@ -4,7 +4,7 @@
 from typing import Literal
 
 from cool_seq_tool.schemas import AnnotationLayer
-from ga4gh.core import domain_models
+from ga4gh.core.models import MappableConcept
 from pydantic import BaseModel, StrictInt, StrictStr
 
 from variation.schemas.app_schemas import AmbiguousRegexType
@@ -255,4 +255,4 @@ class GeneToken(Token):
 
     matched_value: StrictStr
     token_type: Literal[TokenType.GENE] = TokenType.GENE
-    gene: domain_models.Gene | None = None
+    gene: MappableConcept | None = None

src/variation/to_copy_number_variation.py

@@ -42,7 +42,11 @@
 from variation.to_vrs import ToVRS
 from variation.tokenize import Tokenize
 from variation.translate import Translate
-from variation.utils import get_priority_sequence_location, get_vrs_loc_seq
+from variation.utils import (
+    get_copy_change_concept,
+    get_priority_sequence_location,
+    get_vrs_loc_seq,
+)
 from variation.validate import Validate
 
 VALID_CLASSIFICATION_TYPES = [
@@ -603,7 +607,8 @@ def parsed_to_copy_number(
         else:
             if is_cx:
                 variation = models.CopyNumberChange(
-                    location=seq_loc, copyChange=request_body.copy_change
+                    location=seq_loc,
+                    copyChange=get_copy_change_concept(request_body.copy_change),
                 )
                 variation.id = ga4gh_identify(variation)
             else:
@@ -715,7 +720,9 @@ def amplification_to_cx_var(
                     vrs_location.id = ga4gh_identify(vrs_location)
                     vrs_cx = models.CopyNumberChange(
                         location=vrs_location,
-                        copyChange=models.CopyChange.EFO_0030072.value,
+                        copyChange=get_copy_change_concept(
+                            models.CopyChange.EFO_0030072
+                        ),
                     )
                     vrs_cx.id = ga4gh_identify(vrs_cx)
                     variation = models.CopyNumberChange(

src/variation/translators/amplification.py

@@ -9,7 +9,7 @@
 from variation.schemas.translation_response_schema import TranslationResult
 from variation.schemas.validation_response_schema import ValidationResult
 from variation.translators.translator import Translator
-from variation.utils import get_priority_sequence_location
+from variation.utils import get_copy_change_concept, get_priority_sequence_location
 
 
 class Amplification(Translator):
@@ -52,7 +52,7 @@ async def translate(
         if priority_seq_loc:
             vrs_cx = models.CopyNumberChange(
                 location=models.SequenceLocation(**priority_seq_loc),
-                copyChange=models.CopyChange.EFO_0030072,
+                copyChange=get_copy_change_concept(models.CopyChange.EFO_0030072),
             )
             vrs_cx.id = ga4gh_identify(vrs_cx)
             vrs_cx = vrs_cx.model_dump(exclude_none=True)

src/variation/translators/translator.py

@@ -6,7 +6,7 @@
 from cool_seq_tool.mappers import ManeTranscript
 from cool_seq_tool.schemas import AnnotationLayer, ManeGeneData, ResidueMode
 from cool_seq_tool.sources import UtaDatabase
-from ga4gh.core import entity_models
+from ga4gh.core.models import Extension
 from ga4gh.vrs import models
 
 from variation.hgvs_dup_del_mode import HGVSDupDelMode
@@ -258,7 +258,7 @@ async def get_p_or_cdna_translation_result(
     @staticmethod
     def _mane_gene_extensions(
         mane_genes: list[ManeGeneData],
-    ) -> list[entity_models.Extension] | None:
+    ) -> list[Extension] | None:
         """Transform mane genes to list of extensions
 
         This is only used in Genomic translators
@@ -270,7 +270,7 @@ def _mane_gene_extensions(
         mane_genes_exts = None
         if mane_genes:
             mane_genes_exts = [
-                entity_models.Extension(
+                Extension(
                     name="mane_genes",
                     value=mane_genes,
                 )

src/variation/utils.py

@@ -8,7 +8,7 @@
 from bioutils.sequences import aa3_to_aa1 as _aa3_to_aa1
 from cool_seq_tool.handlers import SeqRepoAccess
 from cool_seq_tool.schemas import ResidueMode
-from ga4gh.core import domain_models
+from ga4gh.core.models import Coding, ConceptMapping, MappableConcept, Relation
 from ga4gh.vrs import models
 
 from variation.schemas.app_schemas import AmbiguousRegexType
@@ -67,12 +67,12 @@ def _get_priority_sequence_location(
 
 
 def get_priority_sequence_location(
-    gene: domain_models.Gene, seqrepo_access: SeqRepoAccess
+    gene: MappableConcept, seqrepo_access: SeqRepoAccess
 ) -> dict | None:
     """Get prioritized sequence location from a gene
     Will prioritize NCBI and then Ensembl. GRCh38 will be chosen over GRCh37.
 
-    :param gene: GA4GH Core Gene
+    :param gene: Mappable Concept containing gene information
     :param seqrepo_access: Client to access seqrepo
     :return: Prioritized sequence location represented as a dictionary if found
     """
@@ -237,3 +237,20 @@ def get_vrs_loc_seq(
     else:
         ref = None
     return ref or None  # get_reference_sequence can return empty str
+
+
+def get_copy_change_concept(efo_code: models.CopyChange) -> MappableConcept:
+    """Get mappable concept for EFO code with exactMatch relation
+
+    :param efo_code: EFO code represented as a CURIE
+    :return: Mappable concept for EFO code with exactMatch relation
+    """
+    return MappableConcept(
+        primaryCode=efo_code,
+        mappings=[
+            ConceptMapping(
+                relation=Relation.EXACT_MATCH,
+                coding=Coding(code=efo_code, system="https://www.ebi.ac.uk/efo/"),
+            )
+        ],
+    )

src/variation/vrs_representation.py

@@ -2,7 +2,8 @@
 
 from cool_seq_tool.handlers import SeqRepoAccess
 from cool_seq_tool.schemas import AnnotationLayer, ResidueMode
-from ga4gh.core import entity_models, ga4gh_identify
+from ga4gh.core import ga4gh_identify
+from ga4gh.core.models import Extension
 from ga4gh.vrs import models, normalize
 from pydantic import ValidationError
 
@@ -97,7 +98,7 @@ def vrs_allele(
         sstate: models.LiteralSequenceExpression | models.ReferenceLengthExpression,
         alt_type: AltType,
         errors: list[str],
-        extensions: list[entity_models.Extension] | None = None,
+        extensions: list[Extension] | None = None,
     ) -> dict | None:
         """Create a VRS Allele object.
 
@@ -154,7 +155,7 @@ def to_vrs_allele(
         cds_start: int | None = None,
         alt: str | None = None,
         residue_mode: ResidueMode = ResidueMode.RESIDUE,
-        extensions: list[entity_models.Extension] | None = None,
+        extensions: list[Extension] | None = None,
     ) -> dict | None:
         """Translate accession and position to VRS Allele Object.