fixed error due to ncbi replacing superkingdom with domain column

Author: zyxue

.gitignore

@@ -6,3 +6,4 @@ __pycache__/
 dist
 htmlcov/
 ncbitax2lin.egg-info/
+build

ncbitax2lin/__init__.py

@@ -1,3 +1,3 @@
 """__init__.py for this project"""
 
-__version__ = "2.0.2"
+__version__ = "2.4.1"

ncbitax2lin/data_io.py

@@ -88,9 +88,17 @@ def read_names_and_nodes(names_file: str, nodes_file: str) -> pd.DataFrame:
 
 def write_lineages_to_disk(df_lineages: pd.DataFrame, output_path: str) -> None:
     """Gzip lineages and write them to disk"""
+    # superkingdom has been renamed to domain in
+    # https://ncbiinsights.ncbi.nlm.nih.gov/2024/06/04/changes-ncbi-taxonomy-classifications/
+    domain_col = "domain"
+
+    # For backwards compatibility with older taxdumps.
+    if "superkingdom" in df_lineages:
+        domain_col = "superkingdom"
+
     cols = [
         "tax_id",
-        "superkingdom",
+        domain_col,
         "phylum",
         "class",
         "order",

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "ncbitax2lin"
-version = "2.3.1"
+version = "2.4.1"
 description = "A tool that converts NCBI taxonomy dump into lineages"
 authors = ["Zhuyi Xue <[email protected]>"]
 readme = "README.md"