Merge branch 'main' of github.com:broadinstitute/seqr-loading-pipelines into dev

Author: bpblanken

v03_pipeline/lib/methods/sex_check.py

@@ -13,6 +13,7 @@
 
 AMBIGUOUS_THRESHOLD_PERC: float = 0.01  # Fraction of samples identified as "ambiguous_sex" above which an error will be thrown.
 AAF_THRESHOLD: float = 0.05  # Alternate allele frequency threshold for `hl.impute_sex`.
+BIALLELIC: int = 2
 XX_FSTAT_THRESHOLD: float = (
     0.5  # F-stat threshold below which a sample will be called XX
 )
@@ -24,7 +25,9 @@
 def call_sex(mt: hl.MatrixTable) -> hl.Table:
     # Filter to SNVs and biallelics
     # NB: We should already have filtered biallelics, but just in case.
-    mt = mt.filter_rows(hl.is_snp(mt.alleles[0], mt.alleles[1]))
+    mt = mt.filter_rows(
+        (hl.len(mt.alleles) == BIALLELIC) & hl.is_snp(mt.alleles[0], mt.alleles[1]),
+    )
 
     # Filter to PASS variants only (variants with empty or missing filter set)
     mt = mt.filter_rows(

v03_pipeline/lib/tasks/update_variant_annotations_table_with_new_samples.py

@@ -188,6 +188,13 @@ def update_table(self, ht: hl.Table) -> hl.Table:
                 self.project_pedigree_paths,
             )
         ]
+
+        # Drop any fields potentially unshared/unused by the annotations.
+        for i, callset_ht in enumerate(callset_hts):
+            for row_field in self.dataset_type.optional_row_fields:
+                if hasattr(callset_ht, row_field):
+                    callset_hts[i] = callset_ht.drop(row_field)
+
         callset_ht = functools.reduce(
             (lambda ht1, ht2: ht1.union(ht2, unify=True)),
             callset_hts,

v03_pipeline/lib/tasks/write_project_family_tables.py

@@ -38,7 +38,7 @@ def __init__(self, *args, **kwargs):
         self.dynamic_write_family_table_tasks = set()
 
     def complete(self) -> bool:
-        return len(self.dynamic_write_family_table_tasks) > 1 and all(
+        return len(self.dynamic_write_family_table_tasks) >= 1 and all(
             write_family_table_task.complete()
             for write_family_table_task in self.dynamic_write_family_table_tasks
         )