Merge pipeline

Author: bpblanken

v03_pipeline/lib/annotations/snv_indel.py

@@ -13,8 +13,8 @@ def AB(mt: hl.MatrixTable, **_: Any) -> hl.Expression:  # noqa: N802
     return hl.bind(
         lambda total: hl.if_else(
             (is_called) & (total != 0) & (hl.len(mt.AD) > 1),
-            hl.float(mt.AD[1] / total),
-            hl.missing(hl.tfloat),
+            hl.float32(mt.AD[1] / total),
+            hl.missing(hl.tfloat32),
         ),
         hl.sum(mt.AD),
     )
@@ -24,8 +24,8 @@ def DP(mt: hl.MatrixTable, **_: Any) -> hl.Expression:  # noqa: N802
     is_called = hl.is_defined(mt.GT)
     return hl.if_else(
         is_called & hl.is_defined(mt.AD),
-        hl.int(hl.min(hl.sum(mt.AD), 32000)),
-        hl.missing(hl.tint),
+        hl.int32(hl.min(hl.sum(mt.AD), 32000)),
+        hl.missing(hl.tint32),
     )
 
 

v03_pipeline/lib/misc/math.py

@@ -0,0 +1,5 @@
+def constrain(number: int | float, lower_bound: int | float, upper_bound: int | float) -> int:
+    if lower_bound > upper_bound:
+        msg = 'Lower bound should be less than or equal to upper bound'
+        raise ValueError(msg)
+    return max(lower_bound, min(number, upper_bound))

v03_pipeline/lib/misc/math_test.py

@@ -0,0 +1,13 @@
+import unittest
+
+from v03_pipeline.lib.misc.math import constrain
+
+
+class TestConstrainFunction(unittest.TestCase):
+
+    def test_constrain(self):
+        self.assertEqual(constrain(5, 0, 10), 5)
+        self.assertEqual(constrain(-3, 0, 10), 0)
+        self.assertEqual(constrain(15, 0, 10), 10)
+        with self.assertRaises(ValueError):
+            constrain(5, 10, 0)

v03_pipeline/lib/tasks/update_project_table_test.py

@@ -81,7 +81,7 @@ def test_update_project_table_task(self) -> None:
                     entries=[
                         hl.Struct(
                             GQ=30,
-                            AB=0.3333333333333333,
+                            AB=hl.eval(hl.float32(0.3333333333333333)),
                             DP=3,
                             GT=hl.Call(alleles=[0, 1], phased=False),
                         ),
@@ -93,7 +93,7 @@ def test_update_project_table_task(self) -> None:
                         ),
                         hl.Struct(
                             GQ=61,
-                            AB=0.6,
+                            AB=hl.eval(hl.float32(0.6)),
                             DP=5,
                             GT=hl.Call(alleles=[0, 1], phased=False),
                         ),

v03_pipeline/lib/tasks/update_variant_annotations_table_with_new_samples.py

@@ -6,6 +6,7 @@
 
 from v03_pipeline.lib.annotations.enums import annotate_enums
 from v03_pipeline.lib.annotations.fields import get_fields
+from v03_pipeline.lib.misc.math import constrain
 from v03_pipeline.lib.misc.util import callset_project_pairs
 from v03_pipeline.lib.model import ReferenceDatasetCollection
 from v03_pipeline.lib.paths import (
@@ -26,7 +27,7 @@
 from v03_pipeline.lib.vep import run_vep
 
 GENCODE_RELEASE = 42
-VARIANTS_PER_VEP_PARTITION = 20e3
+VARIANTS_PER_VEP_PARTITION = 1e3
 
 
 class UpdateVariantAnnotationsTableWithNewSamplesTask(BaseVariantAnnotationsTableTask):
@@ -204,7 +205,7 @@ def update_table(self, ht: hl.Table) -> hl.Table:
         new_variants_ht = callset_ht.anti_join(ht)
         new_variants_count = new_variants_ht.count()
         new_variants_ht = new_variants_ht.repartition(
-            max(math.ceil(new_variants_count / VARIANTS_PER_VEP_PARTITION), 1),
+            constrain(math.ceil(new_variants_count / VARIANTS_PER_VEP_PARTITION), 10, 10000),
         )
         new_variants_ht = run_vep(
             new_variants_ht,

v03_pipeline/lib/tasks/write_family_table_test.py

@@ -71,7 +71,7 @@ def test_snv_write_family_table_task(self) -> None:
                 [
                     hl.Struct(
                         GQ=30,
-                        AB=0.3333333333333333,
+                        AB=hl.eval(hl.float32(0.3333333333333333)),
                         DP=3,
                         GT=hl.Call(alleles=[0, 1], phased=False),
                     ),
@@ -83,7 +83,7 @@ def test_snv_write_family_table_task(self) -> None:
                     ),
                     hl.Struct(
                         GQ=61,
-                        AB=0.6,
+                        AB=hl.eval(hl.float32(0.6)),
                         DP=5,
                         GT=hl.Call(alleles=[0, 1], phased=False),
                     ),
@@ -97,7 +97,7 @@ def test_snv_write_family_table_task(self) -> None:
                     ),
                     hl.Struct(
                         GQ=99,
-                        AB=0.5283018867924528,
+                        AB=hl.eval(hl.float32(0.5283018867924528)),
                         DP=53,
                         GT=hl.Call(alleles=[0, 1], phased=False),
                     ),
@@ -123,7 +123,7 @@ def test_snv_write_family_table_task(self) -> None:
                     ),
                     hl.Struct(
                         GQ=99,
-                        AB=0.4146341463414634,
+                        AB=hl.eval(hl.float32(0.4146341463414634)),
                         DP=41,
                         GT=hl.Call(alleles=[0, 1], phased=False),
                     ),