update naming

scripts/variantstore/wdl/extract/create_ranges_cohort_extract_data_table.py

@@ -26,7 +26,7 @@
 output_table_prefix = str(uuid.uuid4()).split("-")[0]
 print(f"running with prefix {output_table_prefix}")
 
-PET_VET_TABLE_COUNT = -1
+REF_VET_TABLE_COUNT = -1
 client = None
 
 EXTRACT_SAMPLE_TABLE = f"{output_table_prefix}_sample_names"
@@ -45,7 +45,7 @@ def dump_job_stats():
   print(" Total GBs billed ", total/(1024 * 1024 * 1024), " GBs")
 
 def get_partition_range(i):
-  if i < 1 or i > PET_VET_TABLE_COUNT:
+  if i < 1 or i > REF_VET_TABLE_COUNT:
     raise ValueError(f"out of partition range")
 
   return { 'start': (i-1)*SAMPLES_PER_PARTITION + 1, 'end': i*SAMPLES_PER_PARTITION }
@@ -152,7 +152,7 @@ def get_ref_subselect(fq_ref_table, samples, id):
           f" `{fq_ref_table}` WHERE sample_id IN ({sample_stanza})), "
     return sql
 
-  for i in range(1, PET_VET_TABLE_COUNT+1):
+  for i in range(1, REF_VET_TABLE_COUNT+1):
     partition_samples = get_samples_for_partition(sample_ids, i)  #sample ids for the partition
 
     if len(partition_samples) > 0:
@@ -185,7 +185,7 @@ def get_ref_subselect(fq_vet_table, samples, id):
           f" `{fq_vet_table}` WHERE sample_id IN ({sample_stanza})), "
     return sql
 
-  for i in range(1, PET_VET_TABLE_COUNT+1):
+  for i in range(1, REF_VET_TABLE_COUNT+1):
     partition_samples = get_samples_for_partition(sample_ids, i)  #sample ids for the partition
 
     if len(partition_samples) > 0:
@@ -265,16 +265,16 @@ def make_extract_table(control_samples,
 
     ## TODO -- provide a cmdline arg to override this (so we can simulate smaller datasets)
 
-    global PET_VET_TABLE_COUNT ## TODO why are we using PET here?
-    PET_VET_TABLE_COUNT = max_tables
+    global REF_VET_TABLE_COUNT ## TODO why are we using PET here?
+    REF_VET_TABLE_COUNT = max_tables
 
     global TEMP_TABLE_TTL_HOURS
     TEMP_TABLE_TTL_HOURS = temp_table_ttl_hours
 
     global TEMP_TABLE_TTL
     TEMP_TABLE_TTL = f" OPTIONS( expiration_timestamp=TIMESTAMP_ADD(CURRENT_TIMESTAMP(), INTERVAL {TEMP_TABLE_TTL_HOURS} HOUR)) "
 
-    print(f"Using {PET_VET_TABLE_COUNT} tables in {fq_ranges_dataset}...")
+    print(f"Using {REF_VET_TABLE_COUNT} tables in {fq_ranges_dataset}...")
 
     # if we have a file of sample names, load it into a temporary table
     if (sample_names_to_extract):