move endpoint from 'main.py' into 'restapi/vlm.py'

Author: jennifer-bowser

src/anyvlm/main.py

@@ -2,24 +2,21 @@
 
 from collections.abc import AsyncGenerator
 from contextlib import asynccontextmanager
-from enum import Enum
-from http import HTTPStatus
-from typing import Annotated, Union
 
-from fastapi import FastAPI, HTTPException, Query, Request
+from fastapi import FastAPI
 
 from anyvlm import __version__
 from anyvlm.anyvar.base_client import BaseAnyVarClient
 from anyvlm.config import get_config
-from anyvlm.functions.get_caf import get_caf
 from anyvlm.schemas.common import (
     SERVICE_DESCRIPTION,
     ServiceInfo,
     ServiceOrganization,
     ServiceType,
 )
-from anyvlm.schemas.vlm import VlmResponse
-from anyvlm.utils.types import ChromosomeName, GenomicSequence, GrcAssemblyId, UscsAssemblyBuild
+from anyvlm.utils.types import (
+    EndpointTag,
+)
 
 
 def create_anyvar_client(
@@ -63,18 +60,11 @@ async def lifespan(app: FastAPI) -> AsyncGenerator:
 )
 
 
-class _Tag(str, Enum):
-    """Define tag names for endpoints."""
-
-    META = "Meta"
-    SEARCH = "Search"
-
-
 @app.get(
     "/service-info",
     summary="Get basic service information",
     description="Retrieve service metadata, such as versioning and contact info. Structured in conformance with the [GA4GH service info API specification](https://www.ga4gh.org/product/service-info/)",
-    tags=[_Tag.META],
+    tags=[EndpointTag.META],
 )
 def service_info() -> ServiceInfo:
     """Provide service info per GA4GH Service Info spec"""
@@ -83,32 +73,3 @@ def service_info() -> ServiceInfo:
         type=ServiceType(),
         environment=get_config().env,
     )
-
-
-@app.get(
-    "/vlm-query",
-    summary="Provides counts of occurrences of a single sequence variant, broken down by zygosity",
-    description="Provides counts of occurrences of a single sequence variant, broken down by zygosity", #TODO: Update this
-    tags=[_Tag.SEARCH]
-)
-def vlm_query(
-    request: Request,
-    assemblyId: Annotated[GrcAssemblyId | UscsAssemblyBuild, Query(..., description="Genome reference assembly")],
-    referenceName: Annotated[ChromosomeName, Query(..., description="Chromosome with optional 'chr' prefix")],
-    start: Annotated[int, Query(..., description="Variant position")],
-    referenceBases: Annotated[GenomicSequence, Query(..., description="Genomic bases ('T', 'AC', etc.)")],
-    alternateBases: Annotated[GenomicSequence, Query(..., description="Genomic bases ('T', 'AC', etc.)")]
-) -> VlmResponse:
-
-    anyvar_client: BaseAnyVarClient = request.app.state.anyvar_client
-    
-    caf_data = get_caf(
-        anyvar_client,
-        assemblyId,
-        referenceName,
-        start,
-        referenceBases,
-        alternateBases
-    )
-
-    return VlmResponse() #TODO: fill this out. See Issue #16 and Issue #13

src/anyvlm/restapi/vlm.py

@@ -1,6 +1,21 @@
 """Define route(s) for the variant-level matching (VLM) protocol"""
 
 from pathlib import Path
+from typing import Annotated
+
+from fastapi import Query, Request
+
+from anyvlm.anyvar.base_client import BaseAnyVarClient
+from anyvlm.functions.get_caf import get_caf
+from anyvlm.main import app
+from anyvlm.schemas.vlm import VlmResponse
+from anyvlm.utils.types import (
+    ChromosomeName,
+    EndpointTag,
+    GenomicSequence,
+    GrcAssemblyId,
+    UscsAssemblyBuild,
+)
 
 
 def ingest_vcf(vcf_path: Path) -> None:
@@ -9,3 +24,45 @@ def ingest_vcf(vcf_path: Path) -> None:
     :param vcf_path: VCF file location
     """
     raise NotImplementedError
+
+
+@app.get(
+    "/vlm-query",
+    summary="Provides counts of occurrences of a single sequence variant, broken down by zygosity",
+    description="Provides counts of occurrences of a single sequence variant, broken down by zygosity",  # TODO: Update this
+    tags=[EndpointTag.SEARCH],
+)
+def vlm_query(
+    request: Request,
+    assemblyId: Annotated[  # noqa: N803
+        GrcAssemblyId | UscsAssemblyBuild,
+        Query(..., description="Genome reference assembly"),
+    ],
+    referenceName: Annotated[  # noqa: N803
+        ChromosomeName, Query(..., description="Chromosome with optional 'chr' prefix")
+    ],
+    start: Annotated[int, Query(..., description="Variant position")],
+    referenceBases: Annotated[  # noqa: N803
+        GenomicSequence, Query(..., description="Genomic bases ('T', 'AC', etc.)")
+    ],
+    alternateBases: Annotated[  # noqa: N803
+        GenomicSequence, Query(..., description="Genomic bases ('T', 'AC', etc.)")
+    ],
+) -> VlmResponse:
+    """Accept a Variant-Level Matching network request and return a count of occurrences of a single sequence variant, broken down by zygosity.
+
+    :param request: FastAPI `Request` object
+    :param assemblyId: The genome reference assembly. Must be a GRC assembly identifier (e.g., "GRCh38) or a USCS assembly build (e.g., "hg38")
+    :param referenceName: The name of the reference chromosome, with optional 'chr' prefix
+    :param start: The start of the variant's position
+    :param referenceBases: Genomic bases ('T', 'AC', etc.)
+    :param alternateBases: Genomic bases ('T', 'AC', etc.)
+    :return: A VlmResponse object containing cohort allele frequency data. If no matches are found, endpoint will return a status code of 200 with an empty set of results.
+    """
+    anyvar_client: BaseAnyVarClient = request.app.state.anyvar_client
+
+    caf_data = get_caf(  # noqa: F841 - TODO: remove this noqa when endpoint is complete. See Issue #16 and Issue #13.
+        anyvar_client, assemblyId, referenceName, start, referenceBases, alternateBases
+    )
+
+    return VlmResponse()  # TODO: fill this out. See Issue #16 and Issue #13

src/anyvlm/utils/__init__.py

@@ -0,0 +1 @@
+"""Provide utilities."""

src/anyvlm/utils/types.py

@@ -1,23 +1,30 @@
 """Provide helpful type definitions, references, and type-based operations."""
 
-from enum import StrEnum
+from enum import Enum, StrEnum
 from typing import Annotated
 
 from pydantic import BeforeValidator, StringConstraints
 
 
+class EndpointTag(str, Enum):
+    """Define tag names for endpoints."""
+
+    META = "Meta"
+    SEARCH = "Search"
+
+
 class GrcAssemblyId(StrEnum):
     """Supported GRC assembly identifiers"""
 
     GRCH37 = "GRCh37"
     GRCH38 = "GRCh38"
-    
+
 
 class UscsAssemblyBuild(StrEnum):
-	"""Supported USCS assembly builds"""
-    
-	HG38 = "hg38"
-	HG19 = "hg19"
+    """Supported USCS assembly builds"""
+
+    HG38 = "hg38"
+    HG19 = "hg19"
 
 
 GenomicSequence = Annotated[
@@ -34,26 +41,29 @@ def _normalize_chromosome_name(chromosome_name: str) -> str:
     :param chromosome_name: The name of the chromosome to normalize, following the rules stated above.
     :return: The chromosome name, stripped of it's 'chr' prefix if it was added
     """
-    error_message = "Invalid chromosome. Must be 1–22, 'X,' or 'Y,' with optional 'chr' prefix."
+    error_message = (
+        "Invalid chromosome. Must be 1-22, 'X,' or 'Y,' with optional 'chr' prefix."
+    )
 
     # strip the 'chr' prefix if it was included
-    chromosome_name = chromosome_name[3:].upper() if chromosome_name.lower().startswith("chr") else chromosome_name.upper()
-    
+    chromosome_name = (
+        chromosome_name[3:].upper()
+        if chromosome_name.lower().startswith("chr")
+        else chromosome_name.upper()
+    )
+
     # chromosome name must be either an int, or "X" or "Y"
     try:
         int(chromosome_name)
-    except:
-        if not chromosome_name in ["X", "Y"]:
-            raise ValueError(error_message)
-    
+    except ValueError:
+        if chromosome_name not in ["X", "Y"]:
+            raise ValueError(error_message) from None
+
     # if chromosome name is an int, it must be between 1-22
-    if chromosome_name not in range(1, 23): # stop is exclusive so we need to add 1
+    if chromosome_name not in range(1, 23):  # stop is exclusive so we need to add 1
         raise ValueError(error_message)
-    
+
     return chromosome_name
 
 
-ChromosomeName = Annotated[
-    str,
-    BeforeValidator(_normalize_chromosome_name)
-]
+ChromosomeName = Annotated[str, BeforeValidator(_normalize_chromosome_name)]