From cb80d69877af839dcceca4d74d1dc67350bf6e33 Mon Sep 17 00:00:00 2001
From: Darren Wight <darren@serenomica.com>
Date: Thu, 28 Aug 2025 12:08:39 +0200
Subject: [PATCH] feat: code added that updates the NAT2 whitelist in the
 pgx-engine

---
 src/create_new_nat2_whitelst.py | 73 +++++++++++++++++++++++++++++++++
 1 file changed, 73 insertions(+)
 create mode 100644 src/create_new_nat2_whitelst.py

diff --git a/src/create_new_nat2_whitelst.py b/src/create_new_nat2_whitelst.py
new file mode 100644
index 0000000..e3906c0
--- /dev/null
+++ b/src/create_new_nat2_whitelst.py
@@ -0,0 +1,73 @@
+import os
+import json
+import re
+import logging
+from datetime import datetime
+import pandas as pd
+
+
+NAT2_JSON_PATH = "/home/darren/Documents/2_repos/serenomica/pgx-engine/resources/allele_whitelist.json"
+NAT2_LOOKUP = "/home/darren/Documents/3_projects/3_PGx/1_Clean_up/pgx_nat2_issue/data/nat2_look-up-table-v1-1.xlsx"
+
+logging.basicConfig(
+    format="%(asctime)s [%(levelname)s] %(message)s", level=logging.INFO
+)
+
+
+def main(old_white: list[str], ref_allele: str, overwrite: bool = False) -> None:
+    logging.info("Script started...")
+    logging.info(f"Loading lookup and filtering to whitelist ('{NAT2_LOOKUP}')...")
+    lookup = pd.read_excel(NAT2_LOOKUP, skiprows=1)
+
+    lookup_re = re.compile(("|").join([f"\\{o}+" for o in old_white]))
+
+    lookup = lookup.loc[
+        (lookup["Transferred\nYes/No"] == "yes")
+        & (lookup["Legacy \nname"].str.contains(lookup_re))
+    ]
+
+    logging.info(
+        f"Saving copy of filtered whitelist: '{(save_path:=os.path.join(
+            os.path.dirname(NAT2_LOOKUP),
+            f"nat2_whitelist_{datetime.now().strftime('%Y%-m-%d')}.tsv",
+        ))}')..."
+    )
+    lookup.to_csv(
+        save_path,
+        sep="\t",
+        index=False,
+    )
+
+    logging.info(f"Reading whitelist ('{NAT2_JSON_PATH}')...")
+    with open(NAT2_JSON_PATH, "r") as handle:
+        whitelist = json.load(handle)
+    genes_pre = set(whitelist.keys())
+
+    logging.info(f"Updating NAT2...")
+    new_entry_json = {
+        "reference": ref_allele,
+        "alleles": sorted(lookup["PharmVar \nName1"].tolist(), reverse=True),
+    }
+
+    whitelist["NAT2"] = new_entry_json
+    genes_post = set(whitelist.keys())
+
+    # sanity check
+    assert len(genes_pre) == len(genes_post)
+    assert genes_pre == genes_post
+
+    if overwrite:
+        logging.info("Overwrite active, saving new white list...")
+        with open(NAT2_JSON_PATH, "w") as handle:
+            json.dump(whitelist, handle)
+    else:
+        logging.info("Overwrite NOT active, printing json to screen...")
+        print(json.dumps(whitelist, indent=2))
+
+    logging.info("Script completed!")
+
+
+if __name__ == "__main__":
+    old_white = ["*4", "*5", "*6", "*7", "*14"]
+    ref_allele = "*4.001"
+    main(old_white, ref_allele, overwrite=True)