Benb/validate with allele type (#785)

* Bump requirements

* add validation

* format

Author: bpblanken

requirements.in

@@ -1,6 +1,6 @@
 elasticsearch==7.9.1
 google-api-python-client>=1.8.0
-hail==0.2.128
+hail==0.2.130
 luigi>=3.4.0
 gnomad==0.6.4
 google-cloud-storage>=2.14.0

requirements.txt

@@ -129,7 +129,7 @@ google-resumable-media==2.7.0
     # via google-cloud-storage
 googleapis-common-protos==1.61.0
     # via google-api-core
-hail==0.2.128
+hail==0.2.130
     # via -r requirements.in
 hdbscan==0.8.33
     # via gnomad
@@ -202,7 +202,7 @@ numpy==1.26.2
     #   scipy
 oauthlib==3.2.2
     # via requests-oauthlib
-orjson==3.9.11
+orjson==3.9.10
     # via hail
 packaging==23.2
     # via
@@ -230,7 +230,6 @@ protobuf==3.20.2
     # via
     #   google-api-core
     #   googleapis-common-protos
-    #   hail
 ptyprocess==0.7.0
     # via pexpect
 pure-eval==0.2.2
@@ -252,9 +251,7 @@ pygments==2.17.2
     #   ipython
     #   rich
 pyjwt[crypto]==2.8.0
-    # via
-    #   msal
-    #   pyjwt
+    # via msal
 pyparsing==3.1.1
     # via httplib2
 pyspark==3.3.3

v03_pipeline/lib/misc/validation.py

@@ -11,6 +11,19 @@ class SeqrValidationError(Exception):
     pass
 
 
+def validate_allele_type(
+    mt: hl.MatrixTable,
+) -> None:
+    ht = mt.rows()
+    ht = ht.filter(
+        hl.numeric_allele_type(ht.alleles[0], ht.alleles[1])
+        == hl.genetics.allele_type.AlleleType.UNKNOWN,
+    )
+    if ht.count() > 0:
+        msg = f'Alleles with Unknown AlleleType are present in the callset: {ht.alleles.collect()}'
+        raise SeqrValidationError(msg)
+
+
 def validate_no_duplicate_variants(
     mt: hl.MatrixTable,
 ) -> None:

v03_pipeline/lib/misc/validation_test.py

@@ -4,6 +4,7 @@
 
 from v03_pipeline.lib.misc.validation import (
     SeqrValidationError,
+    validate_allele_type,
     validate_expected_contig_frequency,
     validate_imputed_sex_ploidy,
     validate_no_duplicate_variants,
@@ -32,6 +33,44 @@ def _mt_from_contigs(contigs):
 
 
 class ValidationTest(unittest.TestCase):
+    def test_validate_allele_type(self) -> None:
+        mt = hl.MatrixTable.from_parts(
+            rows={
+                'locus': [
+                    hl.Locus(
+                        contig='chr1',
+                        position=1,
+                        reference_genome='GRCh38',
+                    ),
+                    hl.Locus(
+                        contig='chr1',
+                        position=2,
+                        reference_genome='GRCh38',
+                    ),
+                    hl.Locus(
+                        contig='chr1',
+                        position=3,
+                        reference_genome='GRCh38',
+                    ),
+                ],
+                'alleles': [
+                    ['A', 'T'],
+                    # NB: star alleles should pass through this validation just fine,
+                    # but are eventually filtered out upstream.
+                    ['A', '*'],
+                    ['A', '-'],
+                ],
+            },
+            cols={'s': ['sample_1']},
+            entries={'HL': [[0.0], [0.0], [0.0]]},
+        ).key_rows_by('locus', 'alleles')
+        self.assertRaisesRegex(
+            SeqrValidationError,
+            "Alleles with Unknown AlleleType are present in the callset: \\[\\['A', '-'\\]\\]",
+            validate_allele_type,
+            mt,
+        )
+
     def test_validate_imputed_sex_ploidy(self) -> None:
         sex_check_ht = hl.read_table(TEST_SEX_CHECK_1)
         mt = hl.MatrixTable.from_parts(

v03_pipeline/lib/tasks/write_imported_callset.py

@@ -7,6 +7,7 @@
     split_multi_hts,
 )
 from v03_pipeline.lib.misc.validation import (
+    validate_allele_type,
     validate_expected_contig_frequency,
     validate_imputed_sex_ploidy,
     validate_no_duplicate_variants,
@@ -134,6 +135,7 @@ def create_table(self) -> hl.MatrixTable:
                 ),
             )
         if self.validate and self.dataset_type.can_run_validation:
+            validate_allele_type(mt)
             validate_no_duplicate_variants(mt)
             validate_expected_contig_frequency(mt, self.reference_genome)
             coding_and_noncoding_ht = hl.read_table(