feat: return MANE gene(s) in normalize endpoint for genomic queries

initial work for #551

src/variation/hgvs_dup_del_mode.py

@@ -2,7 +2,7 @@
 
 from cool_seq_tool.handlers import SeqRepoAccess
 from cool_seq_tool.schemas import ResidueMode
-from ga4gh.core import ga4gh_identify
+from ga4gh.core import entity_models, ga4gh_identify
 from ga4gh.vrs import models, normalize
 
 from variation.schemas.normalize_response_schema import HGVSDupDelModeOption
@@ -49,6 +49,7 @@ 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,
  ) -> dict | None:
  """Use default characteristics to return a variation.
  If baseline_copies not provided and endpoints are ambiguous - copy_number_change
@@ -65,31 +66,40 @@ def default_mode(
  :param baseline_copies: Baseline copies for Copy Number Count variation
  :param copy_change: copy change for Copy Number Change Variation
  :param alt: Alteration
+ :param extensions: List of extensions for variation
  :raises ValueError: If ``alt_type`` not one of ``DELS_DUPS``.
  :return: VRS Variation object represented as a dict
  """
  _check_supported_alt_type(alt_type)
 
  variation = None
  if not baseline_copies and alt_type in AMBIGUOUS_REGIONS:
- variation = self.copy_number_change_mode(alt_type, location, copy_change)
+ variation = self.copy_number_change_mode(
+ alt_type, location, copy_change, extensions=extensions
+ )
  elif baseline_copies:
- variation = self.copy_number_count_mode(alt_type, location, baseline_copies)
+ variation = self.copy_number_count_mode(
+ alt_type, location, baseline_copies, extensions=extensions
+ )
  else:
- variation = self.allele_mode(location, alt_type, vrs_seq_loc_ac, alt)
+ variation = self.allele_mode(
+ location, alt_type, vrs_seq_loc_ac, alt, extensions=extensions
+ )
  return variation
 
  def copy_number_count_mode(
  self,
  alt_type: AltType,
  location: dict,
  baseline_copies: int,
+ extensions: list[entity_models.Extension] | None = None,
  ) -> dict:
  """Return a VRS Copy Number Variation.
 
  :param alt_type: The type of alteration. Must be one of ``DELS_DUPS``.
  :param location: VRS SequenceLocation
  :param baseline_copies: Baseline copies number
+ :param extensions: List of extensions for variation
  :raises ValueError: If ``alt_type`` not one of ``DELS_DUPS``.
  :return: VRS Copy Number object represented as a dict
  """
@@ -98,7 +108,9 @@ def copy_number_count_mode(
  copies = baseline_copies - 1 if alt_type in DELS else baseline_copies + 1
  seq_loc = models.SequenceLocation(**location)
  seq_loc.id = ga4gh_identify(seq_loc)
- cn = models.CopyNumberCount(copies=copies, location=seq_loc)
+ cn = models.CopyNumberCount(
+ copies=copies, location=seq_loc, extensions=extensions
+ )
  cn.id = ga4gh_identify(cn)
  return cn.model_dump(exclude_none=True)
 
@@ -107,12 +119,14 @@ def copy_number_change_mode(
  alt_type: AltType,
  location: dict,
  copy_change: models.CopyChange | None = None,
+ extensions: list[entity_models.Extension] | None = None,
  ) -> dict:
  """Return copy number change variation
 
  :param alt_type: The type of alteration. Must be one of ``DELS_DUPS``.
  :param location: VRS SequenceLocation
  :param copy_change: The copy change
+ :param extensions: List of extensions for variation
  :raises ValueError: If ``alt_type`` not one of ``DELS_DUPS``.
  :return: Copy Number Change variation as a dict
  """
@@ -127,7 +141,9 @@ 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)
+ cx = models.CopyNumberChange(
+ location=seq_loc, copyChange=copy_change, extensions=extensions
+ )
  cx.id = ga4gh_identify(cx)
  return cx.model_dump(exclude_none=True)
 
@@ -137,6 +153,7 @@ def allele_mode(
  alt_type: AltType,
  vrs_seq_loc_ac: str,
  alt: str,
+ extensions: list[entity_models.Extension] | None = None,
  ) -> dict | None:
  """Return a VRS Allele with a normalized LiteralSequenceExpression or
  ReferenceLengthExpression.
@@ -145,6 +162,7 @@ def allele_mode(
  :param alt_type: Alteration type
  :param vrs_seq_loc_ac: Accession used in VRS Sequence Location
  :param alt: Alteration
+ :param extensions: List of extensions for variation
  :return: VRS Allele object represented as a dict
  """
  if alt_type in AMBIGUOUS_REGIONS:
@@ -168,6 +186,7 @@ def allele_mode(
  allele = models.Allele(
  location=models.SequenceLocation(**location),
  state=models.LiteralSequenceExpression(sequence=state),
+ extensions=extensions,
  )
 
  try:
@@ -189,6 +208,7 @@ 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,
  ) -> dict:
  """Interpret variation using HGVSDupDelMode
 
@@ -201,6 +221,7 @@ def interpret_variation(
  :param baseline_copies: Baseline copies number
  :param copy_change: The copy change
  :param alt: The alteration
+ :param extensions: List of extensions for variation
  :return: VRS Variation object
  """
  variation = None
@@ -212,21 +233,24 @@ def interpret_variation(
  baseline_copies=baseline_copies,
  copy_change=copy_change,
  alt=alt,
+ extensions=extensions,
  )
  elif hgvs_dup_del_mode == HGVSDupDelModeOption.ALLELE:
- variation = self.allele_mode(location, alt_type, vrs_seq_loc_ac, alt)
+ variation = self.allele_mode(
+ location, alt_type, vrs_seq_loc_ac, alt, extensions=extensions
+ )
  elif hgvs_dup_del_mode == HGVSDupDelModeOption.COPY_NUMBER_COUNT:
  if baseline_copies:
  variation = self.copy_number_count_mode(
- alt_type, location, baseline_copies
+ alt_type, location, baseline_copies, extensions=extensions
  )
  else:
  errors.append(
  "`baseline_copies` must be provided for Copy Number Count Variation"
  )
  elif hgvs_dup_del_mode == HGVSDupDelModeOption.COPY_NUMBER_CHANGE:
  variation = self.copy_number_change_mode(
- alt_type, location, copy_change=copy_change
+ alt_type, location, copy_change=copy_change, extensions=extensions
  )
 
  if not variation:

src/variation/translators/genomic_del_dup_base.py

@@ -2,7 +2,7 @@
 
 from typing import NamedTuple
 
-from cool_seq_tool.schemas import ResidueMode
+from cool_seq_tool.schemas import ManeGeneData, ResidueMode
 from ga4gh.vrs import models
 from pydantic import StrictInt, StrictStr, ValidationError
 
@@ -30,6 +30,7 @@ class DelDupData(NamedTuple):
  ac: StrictStr
  pos0: StrictInt
  pos1: StrictInt | None
+ mane_genes: list[ManeGeneData] | None
 
 
 class GenomicDelDupTranslator(Translator):
@@ -51,32 +52,34 @@ async def get_grch38_data(
  :param ac: Genomic RefSeq accession
  :return: Data on GRCh38 assembly if successful liftover. Else, `None`
  """
- pos0, pos1, new_ac = None, None, None
+ pos0, pos1, new_ac, mane_genes = None, None, None, None
 
  if classification.pos1:
  # `g_to_grch38` return inter-residue, but we want residue here
  # so we increment start by 1
  grch38_pos = await self.mane_transcript.g_to_grch38(
- ac, classification.pos0 + 1, classification.pos1
+ ac, classification.pos0 + 1, classification.pos1, get_mane_genes=True
  )
  if grch38_pos:
  pos0, pos1 = grch38_pos.pos
  new_ac = grch38_pos.ac
+ mane_genes = grch38_pos.mane_genes
  else:
  # `g_to_grch38` return inter-residue, but we want residue here
  # so we increment start by 1
  grch38_pos = await self.mane_transcript.g_to_grch38(
- ac, classification.pos0 + 1, classification.pos0
+ ac, classification.pos0 + 1, classification.pos0, get_mane_genes=True
  )
  if grch38_pos:
  pos0, _ = grch38_pos.pos
  new_ac = grch38_pos.ac
+ mane_genes = grch38_pos.mane_genes
 
  if not new_ac:
  errors.append(f"Unable to find a GRCh38 accession for: {ac}")
 
  try:
- data = DelDupData(ac=new_ac, pos0=pos0, pos1=pos1)
+ data = DelDupData(ac=new_ac, pos0=pos0, pos1=pos1, mane_genes=mane_genes)
  except ValidationError:
  data = None
  return data
@@ -114,6 +117,7 @@ async def translate(
  vrs_variation = None
  vrs_seq_loc_ac_status = VrsSeqLocAcStatus.NA
  residue_mode = ResidueMode.RESIDUE
+ mane_genes = None
 
  if do_liftover or endpoint_name == Endpoint.NORMALIZE:
  errors = []
@@ -122,15 +126,16 @@ async def translate(
  warnings.append(w)
  return None
 
+ grch38_data = await self.get_grch38_data(
+ classification, errors, validation_result.accession
+ )
+ if assembly == ClinVarAssembly.GRCH37 and errors:
+ warnings += errors
+ return None
+ mane_genes = grch38_data.mane_genes
+
  # assembly is either 37 or 38
  if assembly == ClinVarAssembly.GRCH37:
- grch38_data = await self.get_grch38_data(
- classification, errors, validation_result.accession
- )
- if errors:
- warnings += errors
- return None
-
  pos0 = grch38_data.pos0 - 1
  if grch38_data.pos1 is None:
  pos1 = grch38_data.pos0
@@ -158,7 +163,9 @@ async def translate(
  pos0 = classification.pos0
  pos1 = classification.pos1
  ac = validation_result.accession
- grch38_data = DelDupData(ac=ac, pos0=pos0, pos1=pos1)
+ grch38_data = DelDupData(
+ ac=ac, pos0=pos0, pos1=pos1, mane_genes=mane_genes
+ )
 
  assembly = ClinVarAssembly.GRCH38
  else:
@@ -184,6 +191,7 @@ async def translate(
  ac = grch38_data.ac
  pos0 = grch38_data.pos0 - 1
  pos1 = grch38_data.pos0 if grch38_data.pos1 is None else grch38_data.pos1
+ mane_genes = grch38_data.mane_genes
  residue_mode = ResidueMode.INTER_RESIDUE
  self.is_valid(classification.gene_token, ac, pos0, pos1, errors)
 
@@ -246,6 +254,7 @@ async def translate(
  baseline_copies=baseline_copies,
  copy_change=copy_change,
  alt=alt,
+ extensions=self._mane_gene_extensions(mane_genes),
  )
  elif endpoint_name == Endpoint.HGVS_TO_COPY_NUMBER_COUNT:
  vrs_variation = self.hgvs_dup_del_mode.copy_number_count_mode(

src/variation/translators/genomic_delins.py

@@ -88,9 +88,11 @@ async def translate(
  vrs_seq_loc_ac = mane.refseq
  coord_type = AnnotationLayer.CDNA
  validation_result.classification = classification
+ extensions = None
  else:
  vrs_seq_loc_ac = mane.ac
  coord_type = AnnotationLayer.GENOMIC
+ extensions = self._mane_gene_extensions(mane.mane_genes)
 
  vrs_allele = self.vrs.to_vrs_allele(
  vrs_seq_loc_ac,
@@ -102,6 +104,7 @@ async def translate(
  alt=classification.inserted_sequence,
  cds_start=mane.coding_start_site if gene else None,
  residue_mode=ResidueMode.INTER_RESIDUE,
+ extensions=extensions,
  )
  else:
  vrs_seq_loc_ac = validation_result.accession

src/variation/translators/genomic_insertion.py

@@ -89,9 +89,11 @@ async def translate(
  vrs_seq_loc_ac = mane.refseq
  coord_type = AnnotationLayer.CDNA
  validation_result.classification = classification
+ extensions = None
  else:
  vrs_seq_loc_ac = mane.ac
  coord_type = AnnotationLayer.GENOMIC
+ extensions = self._mane_gene_extensions(mane.mane_genes)
 
  vrs_allele = self.vrs.to_vrs_allele(
  vrs_seq_loc_ac,
@@ -103,6 +105,7 @@ async def translate(
  alt=classification.inserted_sequence,
  cds_start=mane.coding_start_site if gene else None,
  residue_mode=ResidueMode.INTER_RESIDUE,
+ extensions=extensions,
  )
  else:
  vrs_seq_loc_ac = validation_result.accession

src/variation/translators/genomic_reference_agree.py

@@ -87,9 +87,11 @@ async def translate(
  vrs_seq_loc_ac = mane.refseq
  coord_type = AnnotationLayer.CDNA
  validation_result.classification = classification
+ extensions = None
  else:
  vrs_seq_loc_ac = mane.ac
  coord_type = AnnotationLayer.GENOMIC
+ extensions = self._mane_gene_extensions(mane.mane_genes)
 
  vrs_allele = self.vrs.to_vrs_allele(
  vrs_seq_loc_ac,
@@ -100,6 +102,7 @@ async def translate(
  warnings,
  cds_start=mane.coding_start_site if gene else None,
  residue_mode=ResidueMode.INTER_RESIDUE,
+ extensions=extensions,
  )
  else:
  vrs_seq_loc_ac = validation_result.accession

src/variation/translators/genomic_substitution.py

@@ -109,9 +109,11 @@ async def translate(
  vrs_seq_loc_ac = mane.refseq
  coord_type = AnnotationLayer.CDNA
  validation_result.classification = classification
+ extensions = None
  else:
  vrs_seq_loc_ac = mane.ac
  coord_type = AnnotationLayer.GENOMIC
+ extensions = self._mane_gene_extensions(mane.mane_genes)
 
  vrs_allele = self.vrs.to_vrs_allele(
  vrs_seq_loc_ac,
@@ -123,6 +125,7 @@ async def translate(
  alt=classification.alt,
  cds_start=mane.coding_start_site if gene else None,
  residue_mode=ResidueMode.INTER_RESIDUE,
+ extensions=extensions,
  )
  else:
  vrs_seq_loc_ac = validation_result.accession

src/variation/translators/translator.py

@@ -4,8 +4,9 @@
 
 from cool_seq_tool.handlers import SeqRepoAccess
 from cool_seq_tool.mappers import ManeTranscript
-from cool_seq_tool.schemas import AnnotationLayer, ResidueMode
+from cool_seq_tool.schemas import AnnotationLayer, ManeGeneData, ResidueMode
 from cool_seq_tool.sources import UtaDatabase
+from ga4gh.core import entity_models
 from ga4gh.vrs import models
 
 from variation.hgvs_dup_del_mode import HGVSDupDelMode
@@ -253,3 +254,25 @@ async def get_p_or_cdna_translation_result(
  )
 
  return None
+
+ @staticmethod
+ def _mane_gene_extensions(
+ mane_genes: list[ManeGeneData],
+ ) -> list[entity_models.Extension] | None:
+ """Transform mane genes to list of extensions
+
+ This is only used in Genomic translators
+
+ :param mane_genes: Optional list of mane gene data
+ :return: List of extensions containing mane gene data if found. Otherwise,
+ ``None``
+ """
+ mane_genes_exts = None
+ if mane_genes:
+ mane_genes_exts = [
+ entity_models.Extension(
+ name="mane_genes",
+ value=mane_genes,
+ )
+ ]
+ return mane_genes_exts