feat(data-pipelines): add step to remove v4 gene constraint

Author: rileyhgrant

data-pipeline/src/data_pipeline/datasets/gnomad_v4/gnomad_v4_constraint.py

@@ -43,3 +43,9 @@ def prepare_gnomad_v4_constraint(path):
     ds = ds.key_by("transcript_id")
 
     return ds
+
+
+def remove_gnomad_v4_constraint(genes_path):
+    ds = hl.read_table(path)
+    ds = ds.drop("gnomad_constraint")
+    return ds

data-pipeline/src/data_pipeline/pipelines/genes.py

@@ -27,7 +27,10 @@
 )
 from data_pipeline.data_types.gene import reject_par_y_genes
 
-from data_pipeline.datasets.gnomad_v4.gnomad_v4_constraint import prepare_gnomad_v4_constraint
+from data_pipeline.datasets.gnomad_v4.gnomad_v4_constraint import (
+    prepare_gnomad_v4_constraint,
+    remove_gnomad_v4_constraint,
+)
 
 pipeline = Pipeline()
 
@@ -373,6 +376,15 @@ def annotate_with_constraint(genes_path, constraint_path):
     },
 )
 
+pipeline.add_task(
+    "remove_constraint_for_release",
+    remove_gnomad_v4_constraint,
+    f"/{genes_subdir}/genes_grch38_annotate_5_no_constraint",
+    {
+        "genes_path": pipeline.get_task("annotate_grch38_genes_step_5"),
+    },
+)
+
 ###############################################
 # Extract transcripts
 ###############################################