feat: add param to VCFAnnotator to optionally require validation chec…

…ks (#345)

* feat: add param to VCFAnnotator to optionally require validation checks

* Adds `require_validation` parameter. If `True`, validation checks must
  pass in order to return a VRS object. If validation checks fail, a
ValueError will be raised. If `False`, then the VRS object will be
returned even if validation checks fail.

* add new line

* add require_validation as param

* update default

src/ga4gh/vrs/extras/vcf_annotation.py

@@ -102,10 +102,17 @@ class SeqRepoProxyType(str, Enum):
     show_default=True,
     help="Skip VRS computation for REF alleles."
 )
+@click.option(
+    "--require_validation",
+    is_flag=True,
+    default=False,
+    show_default=True,
+    help="Require validation checks to pass in order to return a VRS object"
+)
 def annotate_click(  # pylint: disable=too-many-arguments
     vcf_in: str, vcf_out: Optional[str], vrs_pickle_out: Optional[str],
     vrs_attributes: bool, seqrepo_dp_type: SeqRepoProxyType, seqrepo_root_dir: str,
-    seqrepo_base_url: str, assembly: str, skip_ref: bool
+    seqrepo_base_url: str, assembly: str, skip_ref: bool, require_validation: bool
 ) -> None:
     """Annotate VCF file via click
 
@@ -118,7 +125,11 @@ def annotate_click(  # pylint: disable=too-many-arguments
     msg = f"Annotating {vcf_in} with the VCF Annotator..."
     _logger.info(msg)
     click.echo(msg)
-    annotator.annotate(vcf_in, vcf_out, vrs_pickle_out, vrs_attributes, assembly, (not skip_ref))
+    annotator.annotate(
+        vcf_in, vcf_out=vcf_out, vrs_pickle_out=vrs_pickle_out,
+        vrs_attributes=vrs_attributes, assembly=assembly,
+        compute_for_ref=(not skip_ref), require_validation=require_validation
+    )
     end = timer()
     msg = f"VCF Annotator finished in {(end - start):.5f} seconds"
     _logger.info(msg)
@@ -166,7 +177,8 @@ def __init__(self, seqrepo_dp_type: SeqRepoProxyType = SeqRepoProxyType.LOCAL,
     def annotate(  # pylint: disable=too-many-arguments,too-many-locals
         self, vcf_in: str, vcf_out: Optional[str] = None,
         vrs_pickle_out: Optional[str] = None, vrs_attributes: bool = False,
-        assembly: str = "GRCh38", compute_for_ref: bool = True
+        assembly: str = "GRCh38", compute_for_ref: bool = True,
+        require_validation: bool = True
     ) -> None:
         """Annotates an input VCF file with VRS Allele IDs & creates a pickle file
         containing the vrs object information.
@@ -179,6 +191,9 @@ def annotate(  # pylint: disable=too-many-arguments,too-many-locals
             Only used if `vcf_out` is provided.
         :param str assembly: The assembly used in `vcf_in` data
         :param compute_for_ref: If true, compute VRS IDs for the reference allele
+        :param bool require_validation: If `True` then validation checks must pass in
+            order to return a VRS object. If `False` then VRS object will be returned
+            even if validation checks fail.
         """
         if not any((vcf_out, vrs_pickle_out)):
             raise VCFAnnotatorException(
@@ -228,8 +243,10 @@ def annotate(  # pylint: disable=too-many-arguments,too-many-locals
                 additional_info_fields += [self.VRS_STARTS_FIELD, self.VRS_ENDS_FIELD, self.VRS_STATES_FIELD]
             try:
                 vrs_field_data = self._get_vrs_data(
-                    record, vrs_data, assembly, additional_info_fields, vrs_attributes,
-                    output_pickle, output_vcf, compute_for_ref
+                    record, vrs_data, assembly, additional_info_fields,
+                    vrs_attributes=vrs_attributes, output_pickle=output_pickle,
+                    output_vcf=output_vcf, compute_for_ref=compute_for_ref,
+                    require_validation=require_validation
                 )
             except Exception as ex:
                 _logger.exception("VRS error on %s-%s", record.chrom, record.pos)
@@ -258,7 +275,8 @@ def annotate(  # pylint: disable=too-many-arguments,too-many-locals
     def _get_vrs_object(  # pylint: disable=too-many-arguments,too-many-locals
         self, vcf_coords: str, vrs_data: Dict, vrs_field_data: Dict, assembly: str,
         vrs_data_key: Optional[str] = None, output_pickle: bool = True,
-        output_vcf: bool = False, vrs_attributes: bool = False
+        output_vcf: bool = False, vrs_attributes: bool = False,
+        require_validation: bool = True
     ) -> None:
         """Get VRS Object given `vcf_coords`. `vrs_data` and `vrs_field_data` will
         be mutated.
@@ -278,9 +296,16 @@ def _get_vrs_object(  # pylint: disable=too-many-arguments,too-many-locals
         :param bool vrs_attributes: If `True` will include VRS_Start, VRS_End,
             VRS_State fields in the INFO field. If `False` will not include these fields.
             Only used if `output_vcf` set to `True`.
+        :param bool require_validation: If `True` then validation checks must pass in
+            order to return a VRS object. If `False` then VRS object will be returned
+            even if validation checks fail. Defaults to `True`.
         """
         try:
-            vrs_obj = self.tlr._from_gnomad(vcf_coords, assembly_name=assembly)
+            vrs_obj = self.tlr._from_gnomad(
+                vcf_coords,
+                assembly_name=assembly,
+                require_validation=require_validation
+            )
         except (ValidationError, TranslatorValidationError) as e:
             vrs_obj = None
             _logger.error("ValidationError when translating %s from gnomad: %s", vcf_coords, str(e))
@@ -330,7 +355,7 @@ def _get_vrs_data(  # pylint: disable=too-many-arguments,too-many-locals
         self, record: pysam.VariantRecord, vrs_data: Dict, assembly: str,  # pylint: disable=no-member
         additional_info_fields: List[str], vrs_attributes: bool = False,
         output_pickle: bool = True, output_vcf: bool = True,
-        compute_for_ref: bool = True
+        compute_for_ref: bool = True, require_validation: bool = True
     ) -> Dict:
         """Get VRS data for record's reference and alt alleles.
 
@@ -351,6 +376,10 @@ def _get_vrs_data(  # pylint: disable=too-many-arguments,too-many-locals
             of associated values. If `output_vcf = False`, an empty dictionary will be
             returned.
         :param compute_for_ref: If true, compute VRS IDs for the reference allele
+        :param bool require_validation: If `True` then validation checks must pass in
+            order to return a VRS object. A `ValidationError` will be raised if
+            validation checks fail. If `False` then VRS object will be returned even if
+            validation checks fail. Defaults to `True`.
         """
         vrs_field_data = {field: [] for field in additional_info_fields} if output_vcf else {}
 
@@ -361,7 +390,7 @@ def _get_vrs_data(  # pylint: disable=too-many-arguments,too-many-locals
             self._get_vrs_object(
                 reference_allele, vrs_data, vrs_field_data, assembly,
                 output_pickle=output_pickle, output_vcf=output_vcf,
-                vrs_attributes=vrs_attributes
+                vrs_attributes=vrs_attributes, require_validation=require_validation
             )
 
         # Get VRS data for alts
@@ -378,7 +407,7 @@ def _get_vrs_data(  # pylint: disable=too-many-arguments,too-many-locals
                 self._get_vrs_object(
                     allele, vrs_data, vrs_field_data, assembly, vrs_data_key=data,
                     output_pickle=output_pickle, output_vcf=output_vcf,
-                    vrs_attributes=vrs_attributes
+                    vrs_attributes=vrs_attributes, require_validation=require_validation
                 )
 
         return vrs_field_data

tests/extras/cassettes/test_get_vrs_object_invalid_input.yaml

@@ -23,7 +23,7 @@ interactions:
       Content-Type:
       - application/json
       Date:
-      - Tue, 19 Sep 2023 05:03:24 GMT
+      - Tue, 13 Feb 2024 23:37:55 GMT
       Server:
       - Werkzeug/2.2.2 Python/3.10.4
     status:
@@ -64,7 +64,7 @@ interactions:
       Content-Type:
       - application/json
       Date:
-      - Tue, 19 Sep 2023 05:03:24 GMT
+      - Tue, 13 Feb 2024 23:37:55 GMT
       Server:
       - Werkzeug/2.2.2 Python/3.10.4
     status:
@@ -94,7 +94,78 @@ interactions:
       Content-Type:
       - text/plain; charset=utf-8
       Date:
-      - Tue, 19 Sep 2023 05:03:24 GMT
+      - Tue, 13 Feb 2024 23:37:55 GMT
+      Server:
+      - Werkzeug/2.2.2 Python/3.10.4
+    status:
+      code: 200
+      message: OK
+- request:
+    body: null
+    headers:
+      Accept:
+      - '*/*'
+      Accept-Encoding:
+      - gzip, deflate
+      Connection:
+      - keep-alive
+      User-Agent:
+      - python-requests/2.31.0
+    method: GET
+    uri: http://localhost:5000/seqrepo/1/metadata/ga4gh:SQ.F-LrLMe1SRpfUZHkQmvkVKFEGaoDeHul
+  response:
+    body:
+      string: "{\n  \"added\": \"2016-08-27T21:23:35Z\",\n  \"aliases\": [\n    \"GRCh38:7\",\n
+        \   \"GRCh38:chr7\",\n    \"GRCh38.p1:7\",\n    \"GRCh38.p1:chr7\",\n    \"GRCh38.p10:7\",\n
+        \   \"GRCh38.p10:chr7\",\n    \"GRCh38.p11:7\",\n    \"GRCh38.p11:chr7\",\n
+        \   \"GRCh38.p12:7\",\n    \"GRCh38.p12:chr7\",\n    \"GRCh38.p2:7\",\n    \"GRCh38.p2:chr7\",\n
+        \   \"GRCh38.p3:7\",\n    \"GRCh38.p3:chr7\",\n    \"GRCh38.p4:7\",\n    \"GRCh38.p4:chr7\",\n
+        \   \"GRCh38.p5:7\",\n    \"GRCh38.p5:chr7\",\n    \"GRCh38.p6:7\",\n    \"GRCh38.p6:chr7\",\n
+        \   \"GRCh38.p7:7\",\n    \"GRCh38.p7:chr7\",\n    \"GRCh38.p8:7\",\n    \"GRCh38.p8:chr7\",\n
+        \   \"GRCh38.p9:7\",\n    \"GRCh38.p9:chr7\",\n    \"MD5:cc044cc2256a1141212660fb07b6171e\",\n
+        \   \"NCBI:NC_000007.14\",\n    \"refseq:NC_000007.14\",\n    \"SEGUID:4+JjCcBVhPCr8vdIhUKFycPv8bY\",\n
+        \   \"SHA1:e3e26309c05584f0abf2f748854285c9c3eff1b6\",\n    \"VMC:GS_F-LrLMe1SRpfUZHkQmvkVKFEGaoDeHul\",\n
+        \   \"sha512t24u:F-LrLMe1SRpfUZHkQmvkVKFEGaoDeHul\",\n    \"ga4gh:SQ.F-LrLMe1SRpfUZHkQmvkVKFEGaoDeHul\"\n
+        \ ],\n  \"alphabet\": \"ACGNRSTY\",\n  \"length\": 159345973\n}\n"
+    headers:
+      Connection:
+      - close
+      Content-Length:
+      - '977'
+      Content-Type:
+      - application/json
+      Date:
+      - Tue, 13 Feb 2024 23:37:55 GMT
+      Server:
+      - Werkzeug/2.2.2 Python/3.10.4
+    status:
+      code: 200
+      message: OK
+- request:
+    body: null
+    headers:
+      Accept:
+      - '*/*'
+      Accept-Encoding:
+      - gzip, deflate
+      Connection:
+      - keep-alive
+      User-Agent:
+      - python-requests/2.31.0
+    method: GET
+    uri: http://localhost:5000/seqrepo/1/sequence/ga4gh:SQ.F-LrLMe1SRpfUZHkQmvkVKFEGaoDeHul?start=140753335&end=140753336
+  response:
+    body:
+      string: A
+    headers:
+      Connection:
+      - close
+      Content-Length:
+      - '1'
+      Content-Type:
+      - text/plain; charset=utf-8
+      Date:
+      - Tue, 13 Feb 2024 23:37:55 GMT
       Server:
       - Werkzeug/2.2.2 Python/3.10.4
     status:

tests/extras/test_vcf_annotation.py

@@ -1,8 +1,10 @@
 """Ensure proper functionality of VCFAnnotator"""
 import gzip
 import os
+import re
 
 import pytest
+from ga4gh.vrs.extras.translator import ValidationError
 
 from ga4gh.vrs.extras.vcf_annotation import VCFAnnotator, VCFAnnotatorException
 
@@ -162,3 +164,17 @@ def test_get_vrs_object_invalid_input(vcf_annotator, caplog):
     # No ALT
     vcf_annotator._get_vrs_object("7-140753336-A-.", {}, [], "GRCh38")
     assert "None was returned when translating 7-140753336-A-. from gnomad" in caplog.text
+
+    # Invalid ref, but not requiring validation checks so no error is raised
+    vcf_annotator._get_vrs_object(
+        "7-140753336-G-T", {}, [], "GRCh38", require_validation=False
+    )
+    assert "Expected reference sequence G on GRCh38:7 at positions (140753335, 140753336) but found A" in caplog.text
+
+    # Invalid ref, but requiring validation checks so an error is raised
+    invalid_ref_seq_msg = "Expected reference sequence C on GRCh38:7 at positions (140753335, 140753336) but found A"
+    with pytest.raises(ValidationError, match=re.escape(invalid_ref_seq_msg)):
+        vcf_annotator._get_vrs_object(
+            "7-140753336-C-T", {}, [], "GRCh38", require_validation=True
+        )
+    assert invalid_ref_seq_msg in caplog.text