From b734c786d7cb030aa22912cac0347876174b4d0c Mon Sep 17 00:00:00 2001
From: Jennifer Chang <jennifer.chang.bioinform@gmail.com>
Date: Thu, 18 Jan 2024 07:50:22 -0800
Subject: [PATCH] Refactor post-processing script to be specific to zika strain
 name fixes

This commit refactors the generic post-processing script to better align
with its specific purpose in Zika ingest. The purpose of this script is
to fix zika strain names based on historical modifications from the fauna
repo. In summary the following changes:

* Rename script to fix-zika-strain-names.py to match the purpose
* Add a docstring to the script
* Replace the accession argument with a strain field argument, which is
  the field that needs to be fixed
---
 ...post_process_metadata.py => fix-zika-strain-names.py} | 9 ++++-----
 ingest/rules/transform.smk                               | 3 +--
 2 files changed, 5 insertions(+), 7 deletions(-)
 rename ingest/bin/{post_process_metadata.py => fix-zika-strain-names.py} (85%)

diff --git a/ingest/bin/post_process_metadata.py b/ingest/bin/fix-zika-strain-names.py
similarity index 85%
rename from ingest/bin/post_process_metadata.py
rename to ingest/bin/fix-zika-strain-names.py
index 3c587e5..6c46bf5 100755
--- a/ingest/bin/post_process_metadata.py
+++ b/ingest/bin/fix-zika-strain-names.py
@@ -8,10 +8,10 @@
 
 def parse_args():
     parser = argparse.ArgumentParser(
-        description="Reformat a NCBI Virus metadata.tsv file for a pathogen build."
+        description="Modify zika strain names by referencing historical modifications from the fauna repo."
     )
-    parser.add_argument("--accession-field", default='accession',
-        help="Field from the records to use as the sequence ID in the FASTA file.")
+    parser.add_argument("--strain-field", default='strain',
+        help="Field from the records to use as the strain name to be fixed.")
 
     return parser.parse_args()
 
@@ -48,8 +48,7 @@ def main():
 
     for index, record in enumerate(stdin):
         record = json.loads(record)
-        record["strain"] = _set_strain_name(record)
-        record["authors"] = record["abbr_authors"]
+        record[args.strain_field] = _set_strain_name(record)
         stdout.write(json.dumps(record) + "\n")
 
 
diff --git a/ingest/rules/transform.smk b/ingest/rules/transform.smk
index a0891e5..cc4e917 100644
--- a/ingest/rules/transform.smk
+++ b/ingest/rules/transform.smk
@@ -85,8 +85,7 @@ rule transform:
                 --abbr-authors-field {params.abbr_authors_field} \
             | ./vendored/apply-geolocation-rules \
                 --geolocation-rules {input.all_geolocation_rules} \
-            | ./bin/post_process_metadata.py \
-                --accession-field {params.id_field} \
+            | ./bin/fix-zika-strain-names.py \
             | ./vendored/merge-user-metadata \
                 --annotations {input.annotations} \
                 --id-field {params.annotations_id} \