more error handling fixes, etc

Author: jsstevenson

src/anyvlm/anyvar/base_client.py

@@ -3,6 +3,7 @@
 import abc
 from collections.abc import Iterable
 
+from anyvar.utils.liftover_utils import ReferenceAssembly
 from anyvar.utils.types import VrsVariation
 
 
@@ -22,7 +23,9 @@ class BaseAnyVarClient(abc.ABC):
 
     @abc.abstractmethod
     def put_allele_expressions(
-        self, expressions: Iterable[str], assembly: str = "GRCh38"
+        self,
+        expressions: Iterable[str],
+        assembly: ReferenceAssembly = ReferenceAssembly.GRCH38,
     ) -> list[str | None]:
         """Submit allele expressions to an AnyVar instance and retrieve corresponding VRS IDs
 

src/anyvlm/anyvar/http_client.py

@@ -4,6 +4,7 @@
 from collections.abc import Iterable
 
 import requests
+from anyvar.utils.liftover_utils import ReferenceAssembly
 from anyvar.utils.types import VrsVariation
 from ga4gh.vrs import models
 
@@ -32,7 +33,9 @@ def __init__(
         self.request_timeout = request_timeout
 
     def put_allele_expressions(
-        self, expressions: Iterable[str], assembly: str = "GRCh38"
+        self,
+        expressions: Iterable[str],
+        assembly: ReferenceAssembly = ReferenceAssembly.GRCH38,
     ) -> list[str | None]:
         """Submit allele expressions to an AnyVar instance and retrieve corresponding VRS IDs
 
@@ -47,7 +50,7 @@ def put_allele_expressions(
             url = f"{self.hostname}/variation"
             payload = {
                 "definition": expression,
-                "assembly_name": assembly,
+                "assembly_name": assembly.value,
                 "input_type": "Allele",
             }
             try:

src/anyvlm/anyvar/python_client.py

@@ -6,7 +6,9 @@
 from anyvar import AnyVar
 from anyvar.storage.base_storage import Storage
 from anyvar.translate.translate import TranslationError, Translator
+from anyvar.utils.liftover_utils import ReferenceAssembly
 from anyvar.utils.types import VrsVariation
+from ga4gh.vrs.dataproxy import DataProxyValidationError
 
 from anyvlm.anyvar.base_client import BaseAnyVarClient
 
@@ -25,7 +27,9 @@ def __init__(self, translator: Translator, storage: Storage) -> None:
         self.av = AnyVar(translator, storage)
 
     def put_allele_expressions(
-        self, expressions: Iterable[str], assembly: str = "GRCh38"
+        self,
+        expressions: Iterable[str],
+        assembly: ReferenceAssembly = ReferenceAssembly.GRCH38,
     ) -> list[str | None]:
         """Submit allele expressions to an AnyVar instance and retrieve corresponding VRS IDs
 
@@ -39,12 +43,17 @@ def put_allele_expressions(
             translated_variation = None
             try:
                 translated_variation = self.av.translator.translate_variation(
-                    expression, assembly=assembly
+                    expression, assembly=assembly.value
                 )
+            except DataProxyValidationError:
+                _logger.exception("Found invalid base in expression %s", expression)
             except TranslationError:
                 _logger.exception("Failed to translate expression: %s", expression)
-            self.av.put_objects([translated_variation])  # type: ignore
-            results.append(translated_variation.id)  # type: ignore
+            if translated_variation:
+                self.av.put_objects([translated_variation])  # type: ignore
+                results.append(translated_variation.id)  # type: ignore
+            else:
+                results.append(None)
         return results
 
     def search_by_interval(

src/anyvlm/functions/ingest_vcf.py

@@ -6,6 +6,7 @@
 from pathlib import Path
 
 import pysam
+from anyvar.utils.liftover_utils import ReferenceAssembly
 from ga4gh.va_spec.base import CohortAlleleFrequencyStudyResult
 
 from anyvlm.anyvar.base_client import BaseAnyVarClient
@@ -50,7 +51,11 @@ def _yield_expression_af_batches(
         yield batch
 
 
-def ingest_vcf(vcf_path: Path, av: BaseAnyVarClient, assembly: str = "GRCh38") -> None:
+def ingest_vcf(
+    vcf_path: Path,
+    av: BaseAnyVarClient,
+    assembly: ReferenceAssembly = ReferenceAssembly.GRCH38,
+) -> None:
     """Extract variant and frequency information from a single VCF
 
     Current assumptions (subject to change):
@@ -74,4 +79,4 @@ def ingest_vcf(vcf_path: Path, av: BaseAnyVarClient, assembly: str = "GRCh38") -
         for variant_id, af in zip(variant_ids, afs, strict=True):  # noqa: B007
             if variant_id is None:
                 continue
-            # put af object here
+            # make call to object store method for putting CAF here

tests/unit/anyvar/test_http_client.py

@@ -53,17 +53,6 @@ def test_put_allele_expressions_catch_httperror():
         client.put_allele_expressions(["1-1000000-A-T"])
 
 
-@pytest.mark.vcr
-def test_put_allele_expressions_invalid_assembly(
-    client: HttpAnyVarClient, alleles: dict
-):
-    with pytest.raises(AnyVarClientError):
-        client.put_allele_expressions(
-            alleles["ga4gh:VA.yi7A2l0uIUMaInQaJnHU_B2Cf_OuZRJg"]["vcf_expression"],
-            assembly="GRCh30000",
-        )
-
-
 @pytest.mark.vcr
 def test_search_by_interval(client: HttpAnyVarClient, alleles: dict):
     """Test `search_by_interval` for a couple of basic cases"""

tests/unit/anyvar/test_python_client.py

@@ -42,6 +42,20 @@ def test_put_allele_expressions(client: PythonAnyVarClient, alleles: dict):
         client.put_allele_expressions([allele_fixture["vcf_expression"]])
 
 
+@pytest.mark.vcf
+def test_put_allele_expressions_handle_invalid(
+    client: PythonAnyVarClient, alleles: dict
+):
+    results = client.put_allele_expressions(["Y-2781761-A-C"])  # wrong REF
+    assert results == [None]
+
+    allele_fixture = alleles["ga4gh:VA.yi7A2l0uIUMaInQaJnHU_B2Cf_OuZRJg"]
+    results = client.put_allele_expressions(
+        ["Y-2781761-A-C", allele_fixture["vcf_expression"]]
+    )
+    assert results == [None, allele_fixture["variation"]["id"]]
+
+
 @pytest.mark.vcr
 def test_search_by_interval(populated_client: PythonAnyVarClient, alleles: dict):
     """Test `search_by_interval` for a couple of basic cases"""

tests/unit/functions/test_ingest_vcf.py

@@ -14,3 +14,8 @@ def input_vcf_path(test_data_dir: Path) -> Path:
 @pytest.mark.vcr
 def test_ingest_vcf(input_vcf_path: Path, anyvar_client: HttpAnyVarClient):
     ingest_vcf(input_vcf_path, anyvar_client)
+
+
+# GRCh37 VCF
+# nonexistent path
+# once storage exists, think about how to validate that AFs are stored