Move float conversion around

Author: bpblanken

v03_pipeline/lib/misc/nested_field.py

@@ -12,7 +12,4 @@ def parse_nested_field(t: hl.MatrixTable | hl.Table, fields: str):
             ]
         else:
             expression = expression[field]
-    # Parse float64s into float32s to save space!
-    if expression.dtype == hl.tfloat64:
-        expression = hl.float32(expression)
     return expression

v03_pipeline/lib/model/dataset_type.py

@@ -138,7 +138,7 @@ def row_fields(
                 'lof_genes': hl.tset(hl.tstr),
                 'num_exon': hl.tint32,
                 'sc': hl.tint32,
-                'sf': hl.tfloat32,
+                'sf': hl.tfloat64,
                 'start': hl.tint32,
                 'strvctvre_score': hl.tstr,
                 'svtype': hl.tstr,

v03_pipeline/lib/reference_datasets/misc.py

@@ -17,6 +17,16 @@
 BIALLELIC = 2
 
 
+def compress_floats(ht: hl.Table):
+    # Parse float64s into float32s to save space!
+    return ht.select(
+        **{
+            k: hl.float32(v) if v.dtype == hl.tfloat64 else v
+            for k, v in ht.row_value.items()
+        },
+    )
+
+
 def generate_random_string(length=5):
     """Generates a random string of the specified length."""
     letters = string.ascii_letters + string.digits

v03_pipeline/lib/reference_datasets/reference_dataset.py

@@ -16,6 +16,7 @@
     filter_contigs,
     filter_mito_contigs,
     get_enum_select_fields,
+    compress_floats,
 )
 
 DATASET_TYPES = 'dataset_types'
@@ -115,6 +116,7 @@ def get_ht(
         )
         path = self.path(reference_genome)
         ht = module.get_ht(path, reference_genome)
+        ht = compress_floats(ht)
         enum_selects = get_enum_select_fields(ht, self.enums)
         if enum_selects:
             ht = ht.transmute(**enum_selects)