feat: sample type validation excluded projects

v03_pipeline/lib/misc/validation.py

@@ -4,6 +4,7 @@
 
 from v03_pipeline.lib.model import (
     DatasetType,
+    Env,
     ReferenceGenome,
     SampleType,
 )
@@ -151,12 +152,18 @@ def _validate_field(
 
 def validate_sample_type(
     mt: hl.MatrixTable,
-    coding_and_noncoding_variants_ht: hl.Table,
     reference_genome: ReferenceGenome,
     sample_type: SampleType,
+    project_guids: list[str],
+    coding_and_noncoding_variants_ht: hl.Table,
     sample_type_match_threshold: float = SAMPLE_TYPE_MATCH_THRESHOLD,
     **_: Any,
 ) -> None:
+    if all(
+        project_guid in Env.SAMPLE_TYPE_VALIDATION_EXCLUDED_PROJECTS
+        for project_guid in project_guids
+    ):
+        return
     coding_variants_ht = coding_and_noncoding_variants_ht.filter(
         coding_and_noncoding_variants_ht.coding,
     )

v03_pipeline/lib/misc/validation_test.py

@@ -1,4 +1,5 @@
 import unittest
+from unittest.mock import patch
 
 import hail as hl
 
@@ -325,19 +326,21 @@ def test_validate_sample_type(self) -> None:
         self.assertIsNone(
             validate_sample_type(
                 mt,
-                coding_and_noncoding_variants_ht,
                 ReferenceGenome.GRCh38,
                 SampleType.WGS,
+                ['project_a'],
+                coding_and_noncoding_variants_ht,
             ),
         )
         self.assertRaisesRegex(
             SeqrValidationError,
             'specified as WES but appears to be WGS',
             validate_sample_type,
             mt,
-            coding_and_noncoding_variants_ht,
             ReferenceGenome.GRCh38,
             SampleType.WES,
+            ['project_a'],
+            coding_and_noncoding_variants_ht,
         )
 
         # has coding, but not noncoding now.
@@ -381,19 +384,21 @@ def test_validate_sample_type(self) -> None:
         self.assertIsNone(
             validate_sample_type(
                 mt,
-                coding_and_noncoding_variants_ht,
                 ReferenceGenome.GRCh38,
                 SampleType.WES,
+                ['project_a'],
+                coding_and_noncoding_variants_ht,
             ),
         )
         self.assertRaisesRegex(
             SeqrValidationError,
             'specified as WGS but appears to be WES',
             validate_sample_type,
             mt,
-            coding_and_noncoding_variants_ht,
             ReferenceGenome.GRCh38,
             SampleType.WGS,
+            ['project_a'],
+            coding_and_noncoding_variants_ht,
         )
 
         # has noncoding, but not coding now.
@@ -439,7 +444,21 @@ def test_validate_sample_type(self) -> None:
             'contains noncoding variants but is missing common coding variants',
             validate_sample_type,
             mt,
-            coding_and_noncoding_variants_ht,
             ReferenceGenome.GRCh38,
             SampleType.WGS,
+            ['project_a'],
+            coding_and_noncoding_variants_ht,
         )
+
+        # Validation safe if all projects are to be skipped
+        with patch('v03_pipeline.lib.misc.validation.Env') as mock_env:
+            mock_env.SAMPLE_TYPE_VALIDATION_EXCLUDED_PROJECTS = ['project_a']
+            self.assertIsNone(
+                validate_sample_type(
+                    mt,
+                    ReferenceGenome.GRCh38,
+                    SampleType.WGS,
+                    ['project_a'],
+                    coding_and_noncoding_variants_ht,
+                ),
+            )

v03_pipeline/lib/model/environment.py

@@ -53,6 +53,11 @@
 GCLOUD_ZONE = os.environ.get('GCLOUD_ZONE')
 GCLOUD_REGION = os.environ.get('GCLOUD_REGION')
 PIPELINE_RUNNER_APP_VERSION = os.environ.get('PIPELINE_RUNNER_APP_VERSION', 'latest')
+SAMPLE_TYPE_VALIDATION_EXCLUDED_PROJECTS = tuple(
+    x
+    for x in os.environ.get('SAMPLE_TYPE_VALIDATION_EXCLUDED_PROJECTS', '').split(',')
+    if x
+)
 SEQR_APP_HAIL_SEARCH_DATA_DIR = os.environ.get('SEQR_APP_HAIL_SEARCH_DATA_DIR')
 SEQR_APP_REFERENCE_DATASETS_DIR = os.environ.get('SEQR_APP_REFERENCE_DATASETS_DIR')
 
@@ -79,6 +84,9 @@ class Env:
     PIPELINE_RUNNER_APP_VERSION: str = PIPELINE_RUNNER_APP_VERSION
     PRIVATE_REFERENCE_DATASETS_DIR: str = PRIVATE_REFERENCE_DATASETS_DIR
     REFERENCE_DATASETS_DIR: str = REFERENCE_DATASETS_DIR
+    SAMPLE_TYPE_VALIDATION_EXCLUDED_PROJECTS: tuple[str] = (
+        SAMPLE_TYPE_VALIDATION_EXCLUDED_PROJECTS
+    )
     SEQR_APP_HAIL_SEARCH_DATA_DIR: str | None = SEQR_APP_HAIL_SEARCH_DATA_DIR
     SEQR_APP_REFERENCE_DATASETS_DIR: str | None = SEQR_APP_REFERENCE_DATASETS_DIR
     VEP_REFERENCE_DATASETS_DIR: str = VEP_REFERENCE_DATASETS_DIR