feat!: adding transcript lookup for genes

client/src/components/Utilities/GetCoordinates/GetCoordinates.tsx

@@ -151,8 +151,8 @@ const GetCoordinates: React.FC = () => {
 
   const handleResponse = (coordsResponse: CoordsUtilsResponse) => {
     if (coordsResponse.warnings) {
-      setResults(null);
-      clearWarnings();
+      // setResults(null);
+      // clearWarnings();
       coordsResponse.warnings.forEach((warning) => {
         if (warning.startsWith("Found more than one accession")) {
           setChromosomeText("Complete ID required");

requirements.txt

@@ -18,7 +18,7 @@ charset-normalizer==3.2.0
 click==8.1.6
 coloredlogs==15.0.1
 configparser==6.0.0
-cool-seq-tool==0.1.14.dev0
+cool-seq-tool==0.3.0-dev1
 cssselect==1.2.0
 Cython==3.0.0
 decorator==5.1.1
@@ -27,8 +27,8 @@ fake-useragent==1.1.3
 fastapi==0.100.0
 fusor==0.0.30.dev1
 ga4gh.vrs==0.8.4
-ga4gh.vrsatile.pydantic==0.0.13
-gene-normalizer==0.1.39
+ga4gh.vrsatile.pydantic==0.2.0
+gene-normalizer==0.1.40-dev1
 h11==0.14.0
 hgvs==1.5.4
 humanfriendly==10.0
@@ -55,7 +55,7 @@ prompt-toolkit==3.0.39
 psycopg2==2.9.6
 ptyprocess==0.7.0
 pure-eval==0.2.2
-pydantic==1.10.12
+pydantic==2.4.2
 pyee==8.2.2
 Pygments==2.15.1
 pyliftover==0.4

server/curfu/routers/demo.py

@@ -101,10 +101,10 @@ def clientify_structural_element(
         element_args["nomenclature"] = nm
         element_args["input_type"] = "exon_coords_tx"
         element_args["input_tx"] = element.transcript.split(":")[1]
-        element_args["input_exon_start"] = element.exon_start
-        element_args["input_exon_start_offset"] = element.exon_start_offset
-        element_args["input_exon_end"] = element.exon_end
-        element_args["input_exon_end_offset"] = element.exon_end_offset
+        element_args["input_exon_start"] = str(element.exon_start)
+        element_args["input_exon_start_offset"] = str(element.exon_start_offset)
+        element_args["input_exon_end"] = str(element.exon_end)
+        element_args["input_exon_end_offset"] = str(element.exon_end_offset)
         return ClientTranscriptSegmentElement(**element_args)
     else:
         raise ValueError("Unknown element type provided")

server/curfu/routers/utilities.py

@@ -64,10 +64,9 @@ async def get_transcripts_for_gene(request: Request, gene: str) -> Dict:
     :param str gene: gene term provided by user
     :return: Dict containing transcripts if lookup succeeds, or warnings upon failure
     """
-    cst = CoolSeqTool(db_url=UTA_DB_URL)
-    transcripts = await cst.uta_db.get_transcripts_from_gene(gene)
-    if not transcripts:
-        return {"warnings": [f"No matching transcripts: {gene}"]}
+    transcripts = await request.app.state.fusor.cool_seq_tool.uta_db.get_transcripts(gene)
+    if transcripts.is_empty():
+        return {"warnings": [f"No matching transcripts: {gene}"], "transcripts": []}
     else:
         return {"transcripts": transcripts}
 
@@ -187,7 +186,7 @@ async def get_exon_coords(
             logger.warning(warning)
         return CoordsUtilsResponse(warnings=warnings, coordinates_data=None)
 
-    response = await request.app.state.fusor.cool_seq_tool.genomic_to_transcript_exon_coordinates(  # noqa: E501
+    response = await request.app.state.fusor.cool_seq_tool.ex_g_coords_mapper.genomic_to_transcript_exon_coordinates(  # noqa: E501
         chromosome,
         start=start,
         end=end,

server/curfu/schemas.py

@@ -1,5 +1,6 @@
 """Provide schemas for FastAPI responses."""
 from typing import Dict, List, Literal, Optional, Tuple, Union
+import polars as pl
 
 from cool_seq_tool.schemas import GenomicData
 from fusor.models import (
@@ -45,16 +46,16 @@ class ClientTranscriptSegmentElement(TranscriptSegmentElement, ClientStructuralE
         Literal["genomic_coords_tx"],
         Literal["exon_coords_tx"],
     ]
-    input_tx: Optional[str]
-    input_strand: Optional[Strand]
-    input_gene: Optional[str]
-    input_chr: Optional[str]
-    input_genomic_start: Optional[str]
-    input_genomic_end: Optional[str]
-    input_exon_start: Optional[str]
-    input_exon_start_offset: Optional[str]
-    input_exon_end: Optional[str]
-    input_exon_end_offset: Optional[str]
+    input_tx: Optional[str] = ""
+    input_strand: Optional[Strand] = None
+    input_gene: Optional[str] = ""
+    input_chr: Optional[str] = ""
+    input_genomic_start: Optional[str] = ""
+    input_genomic_end: Optional[str] = ""
+    input_exon_start: Optional[str] = ""
+    input_exon_start_offset: Optional[str] = ""
+    input_exon_end: Optional[str] = ""
+    input_exon_end_offset: Optional[str] = ""
 
 
 class ClientLinkerElement(LinkerElement, ClientStructuralElement):
@@ -66,9 +67,9 @@ class ClientLinkerElement(LinkerElement, ClientStructuralElement):
 class ClientTemplatedSequenceElement(TemplatedSequenceElement, ClientStructuralElement):
     """Templated sequence element used client-side."""
 
-    input_chromosome: Optional[str]
-    input_start: Optional[str]
-    input_end: Optional[str]
+    input_chromosome: Optional[str] = ""
+    input_start: Optional[str] = ""
+    input_end: Optional[str] = ""
 
 
 class ClientGeneElement(GeneElement, ClientStructuralElement):
@@ -112,7 +113,7 @@ class ClientRegulatoryElement(RegulatoryElement):
 class Response(BaseModel):
     """Abstract Response class for defining API response structures."""
 
-    warnings: ResponseWarnings
+    warnings: ResponseWarnings = None
 
     class Config:
         """Configure class"""
@@ -123,28 +124,28 @@ class Config:
 class GeneElementResponse(Response):
     """Response model for gene element construction endoint."""
 
-    element: Optional[GeneElement]
+    element: Optional[GeneElement] = None
 
 
 class TxSegmentElementResponse(Response):
     """Response model for transcript segment element construction endpoint."""
 
-    element: Optional[TranscriptSegmentElement]
+    element: Optional[TranscriptSegmentElement] = None
 
 
 class TemplatedSequenceElementResponse(Response):
     """Response model for transcript segment element construction endpoint."""
 
-    element: Optional[TemplatedSequenceElement]
+    element: Optional[TemplatedSequenceElement] = None
 
 
 class NormalizeGeneResponse(Response):
     """Response model for gene normalization endpoint."""
 
     term: StrictStr
-    concept_id: Optional[CURIE]
-    symbol: Optional[StrictStr]
-    cased: Optional[StrictStr]
+    concept_id: Optional[CURIE] = None
+    symbol: Optional[StrictStr] = ""
+    cased: Optional[StrictStr] = ""
 
 
 class SuggestGeneResponse(Response):
@@ -153,10 +154,10 @@ class SuggestGeneResponse(Response):
     term: StrictStr
     matches_count: int
     # complete term, normalized symbol, normalized concept ID, chromosome ID, strand
-    concept_id: Optional[List[Tuple[str, str, str, str, str]]]
-    symbol: Optional[List[Tuple[str, str, str, str, str]]]
-    prev_symbols: Optional[List[Tuple[str, str, str, str, str]]]
-    aliases: Optional[List[Tuple[str, str, str, str, str]]]
+    concept_id: Optional[List[Tuple[str, str, str, str, str]]] = []
+    symbol: Optional[List[Tuple[str, str, str, str, str]]] = []
+    prev_symbols: Optional[List[Tuple[str, str, str, str, str]]] = []
+    aliases: Optional[List[Tuple[str, str, str, str, str]]] = []
 
 
 class DomainParams(BaseModel):
@@ -172,20 +173,20 @@ class DomainParams(BaseModel):
 class GetDomainResponse(Response):
     """Response model for functional domain constructor endpoint."""
 
-    domain: Optional[FunctionalDomain]
+    domain: Optional[FunctionalDomain] = None
 
 
 class AssociatedDomainResponse(Response):
     """Response model for domain ID autocomplete suggestion endpoint."""
 
     gene_id: StrictStr
-    suggestions: Optional[List[DomainParams]]
+    suggestions: Optional[List[DomainParams]] = []
 
 
 class ValidateFusionResponse(Response):
     """Response model for Fusion validation endpoint."""
 
-    fusion: Optional[Fusion]
+    fusion: Optional[Fusion] = None
 
 
 class ExonCoordsRequest(BaseModel):
@@ -223,9 +224,9 @@ class SequenceIDResponse(Response):
     """Response model for sequence ID retrieval endpoint."""
 
     sequence: StrictStr
-    refseq_id: Optional[StrictStr]
-    ga4gh_id: Optional[StrictStr]
-    aliases: Optional[List[StrictStr]]
+    refseq_id: Optional[StrictStr] = ""
+    ga4gh_id: Optional[StrictStr] = ""
+    aliases: Optional[List[StrictStr]] = []
 
 
 class ManeGeneTranscript(BaseModel):
@@ -242,8 +243,8 @@ class ManeGeneTranscript(BaseModel):
     Ensembl_prot: str
     MANE_status: str
     GRCh38_chr: str
-    chr_start: str
-    chr_end: str
+    chr_start: StrictInt
+    chr_end: StrictInt
     chr_strand: str
 
 
@@ -255,7 +256,7 @@ class GetTranscriptsResponse(Response):
 class GetGeneTranscriptsResponse(Response):
     """Response model for MANE transcript retrieval endpoint."""
 
-    transcripts: Optional[List[str]]
+    transcripts: Optional[pl.DataFrame]
 
 
 class ServiceInfoResponse(Response):
@@ -309,7 +310,7 @@ class ClientAssayedFusion(AssayedFusion):
 class NomenclatureResponse(Response):
     """Response model for regulatory element nomenclature endpoint."""
 
-    nomenclature: Optional[str]
+    nomenclature: Optional[str] = ""
 
 
 class RegulatoryElementResponse(Response):

server/setup.cfg

@@ -36,7 +36,9 @@ dev =
     ruff
     black
     pre-commit
-    gene-normalizer ~= 0.1.39
+    gene-normalizer ~=0.1.40-dev1
+    cool-seq-tool ~=0.3.0-dev1
+    pydantic == 2.*
     pydantic-to-typescript