add task to write relatedness check to tsv (#930)

* add task to write relatedness check to tsv

* fix requirements

* relatedness_check_table_path

Author: jklugherz

v03_pipeline/lib/paths.py

@@ -187,6 +187,22 @@ def relatedness_check_table_path(
     )
 
 
+def relatedness_check_tsv_path(
+    reference_genome: ReferenceGenome,
+    dataset_type: DatasetType,
+    callset_path: str,
+) -> str:
+    return os.path.join(
+        _pipeline_prefix(
+            Env.LOADING_DATASETS_DIR,
+            reference_genome,
+            dataset_type,
+        ),
+        'relatedness_check',
+        f'{hashlib.sha256(callset_path.encode("utf8")).hexdigest()}.tsv',
+    )
+
+
 def remapped_and_subsetted_callset_path(
     reference_genome: ReferenceGenome,
     dataset_type: DatasetType,

v03_pipeline/lib/tasks/write_relatedness_check_tsv.py

@@ -0,0 +1,29 @@
+import hail as hl
+import luigi
+import luigi.util
+
+from v03_pipeline.lib.paths import relatedness_check_tsv_path
+from v03_pipeline.lib.tasks.base.base_loading_run_params import BaseLoadingRunParams
+from v03_pipeline.lib.tasks.files import GCSorLocalTarget
+from v03_pipeline.lib.tasks.write_relatedness_check_table import (
+    WriteRelatednessCheckTableTask,
+)
+
+
+@luigi.util.inherits(BaseLoadingRunParams)
+class WriteRelatednessCheckTsvTask(luigi.Task):
+    def output(self) -> luigi.Target:
+        return GCSorLocalTarget(
+            relatedness_check_tsv_path(
+                self.reference_genome,
+                self.dataset_type,
+                self.callset_path,
+            ),
+        )
+
+    def requires(self):
+        return [self.clone(WriteRelatednessCheckTableTask)]
+
+    def run(self):
+        ht = hl.read_table(self.input()[0].path)
+        ht.export(self.output().path)

v03_pipeline/lib/tasks/write_relatedness_check_tsv_test.py

@@ -0,0 +1,53 @@
+import shutil
+
+import luigi.worker
+
+from v03_pipeline.lib.model import DatasetType, ReferenceGenome, SampleType
+from v03_pipeline.lib.paths import relatedness_check_table_path
+from v03_pipeline.lib.tasks.write_relatedness_check_tsv import (
+    WriteRelatednessCheckTsvTask,
+)
+from v03_pipeline.lib.test.mocked_dataroot_testcase import MockedDatarootTestCase
+
+TEST_RELATEDNESS_CHECK_1 = (
+    'v03_pipeline/var/test/relatedness_check/test_relatedness_check_1.ht'
+)
+TEST_VCF = 'v03_pipeline/var/test/callsets/1kg_30variants.vcf'
+TEST_RUN_ID = 'manual__2024-04-03'
+
+
+class WriteRelatednessCheckTsvTaskTest(MockedDatarootTestCase):
+    def setUp(self) -> None:
+        super().setUp()
+        shutil.copytree(
+            TEST_RELATEDNESS_CHECK_1,
+            relatedness_check_table_path(
+                ReferenceGenome.GRCh38,
+                DatasetType.SNV_INDEL,
+                TEST_VCF,
+            ),
+        )
+
+    def test_write_relatedness_check_tsv_task(
+        self,
+    ) -> None:
+        worker = luigi.worker.Worker()
+        task = WriteRelatednessCheckTsvTask(
+            reference_genome=ReferenceGenome.GRCh38,
+            dataset_type=DatasetType.SNV_INDEL,
+            callset_path=TEST_VCF,
+            run_id=TEST_RUN_ID,
+            sample_type=SampleType.WES,
+        )
+        worker.add(task)
+        worker.run()
+        self.assertTrue(task.complete())
+        with task.output().open('r') as f:
+            lines = f.readlines()
+            expected_lines = [
+                'i\tj\tibd0\tibd1\tibd2\tpi_hat\n',
+                'HG00731_1\tHG00733_1\t0\t1\t0\t5.0000e-01\n',
+                'HG00732_1\tHG00733_1\t0\t1\t0\t5.0000e-01\n',
+            ]
+            for expected_line, actual_line in zip(expected_lines, lines, strict=False):
+                self.assertEqual(expected_line, actual_line)

v03_pipeline/lib/tasks/write_remapped_and_subsetted_callset.py

@@ -17,13 +17,16 @@
 from v03_pipeline.lib.misc.pedigree import parse_pedigree_ht_to_families
 from v03_pipeline.lib.misc.sample_ids import remap_sample_ids, subset_samples
 from v03_pipeline.lib.model.environment import Env
-from v03_pipeline.lib.paths import remapped_and_subsetted_callset_path
+from v03_pipeline.lib.paths import (
+    relatedness_check_table_path,
+    remapped_and_subsetted_callset_path,
+)
 from v03_pipeline.lib.tasks.base.base_loading_run_params import BaseLoadingRunParams
 from v03_pipeline.lib.tasks.base.base_write import BaseWriteTask
 from v03_pipeline.lib.tasks.files import GCSorLocalTarget, RawFileTask
 from v03_pipeline.lib.tasks.validate_callset import ValidateCallsetTask
-from v03_pipeline.lib.tasks.write_relatedness_check_table import (
-    WriteRelatednessCheckTableTask,
+from v03_pipeline.lib.tasks.write_relatedness_check_tsv import (
+    WriteRelatednessCheckTsvTask,
 )
 from v03_pipeline.lib.tasks.write_sex_check_table import WriteSexCheckTableTask
 
@@ -67,7 +70,7 @@ def requires(self) -> list[luigi.Task]:
         ):
             requirements = [
                 *requirements,
-                self.clone(WriteRelatednessCheckTableTask),
+                self.clone(WriteRelatednessCheckTsvTask),
                 self.clone(WriteSexCheckTableTask),
             ]
         return requirements
@@ -101,7 +104,13 @@ def create_table(self) -> hl.MatrixTable:
             and self.dataset_type.check_sex_and_relatedness
             and not self.skip_check_sex_and_relatedness
         ):
-            relatedness_check_ht = hl.read_table(self.input()[2].path)
+            relatedness_check_ht = hl.read_table(
+                relatedness_check_table_path(
+                    self.reference_genome,
+                    self.dataset_type,
+                    self.callset_path,
+                ),
+            )
             sex_check_ht = hl.read_table(self.input()[3].path)
             families_failed_relatedness_check = get_families_failed_relatedness_check(
                 families - families_failed_missing_samples.keys(),