ENH Bundle aro.obo with argNorm

Previously, it was repeatedly loaded from the internet

CHANGELOG.md

@@ -1,5 +1,8 @@
 # Changelog
 
+## Unreleased
+
+- Bundle a specific version of ARO with the package instead of downloading it from the internet (ensures reproducibility)
 
 ## 0.3.0 - 27 April 2024
 

argnorm/drug_categorization.py

@@ -1,8 +1,7 @@
-import pronto
 from typing import List
+from . import lib
 
-# Load the ARO ontology from internet
-ARO = pronto.Ontology.from_obo_library('aro.obo')
+ARO = lib.get_aro_ontology()
 confers_resistance_to_drug_class_rel = ARO.get_relationship('confers_resistance_to_drug_class')
 confers_resistance_to_antibiotic_rel = ARO.get_relationship('confers_resistance_to_antibiotic')
 

argnorm/lib.py

@@ -5,13 +5,30 @@
     pd.options.mode.copy_on_write = True
 except pd.errors.OptionError:
     pass
-import pronto
 
 ORIGINAL_ID_COL = 'Original ID'
 MAPPING_TABLE_ARO_COL = 'ARO'
 TARGET_ARO_COL = 'ARO'
 
 _ROOT = os.path.abspath(os.path.dirname(__file__))
+_ARO = None
+
+def get_aro_ontology():
+    """
+    The ARO ontology used by (bundled with) argNorm
+
+    Returns:
+        ARO (pronto.Ontology): A pronto ontology object with ARO terms.
+    """
+    import pronto
+    import importlib.resources
+    global _ARO
+    if _ARO is None:
+        _ARO = pronto.Ontology(
+                importlib.resources.files('argnorm')
+                            .joinpath('data/aro.obo')
+                            .open('rb'))
+    return _ARO
 
 def get_aro_mapping_table(database):
     """
@@ -66,6 +83,5 @@ def map_to_aro(gene, database):
     else:
         if pd.isna(result):
             return None
-
-        ARO = pronto.Ontology.from_obo_library('aro.obo')
-        return ARO.get(result)
+        ARO = get_aro_ontology()
+        return ARO.get(result)

setup.py

@@ -16,7 +16,11 @@
     packages=['argnorm', 'argnorm.data'],
     include_package_data=True,
     package_dir={'argnorm': 'argnorm' },
-    package_data={'argnorm': ['data/*.tsv', 'data/manual_curation/*.tsv']},
+    package_data={'argnorm': [
+        'data/aro.obo',
+        'data/*.tsv',
+        'data/manual_curation/*.tsv'
+        ]},
     install_requires=open("./requirements.txt", "r").read().splitlines(),
     long_description=open("./README.md", "r").read(),
     long_description_content_type='text/markdown',

tests/test_lib.py

@@ -1,6 +1,6 @@
 import pytest
+from argnorm import lib
 from argnorm.lib import map_to_aro, get_aro_mapping_table
-import pronto
 
 def test_map_to_aro():
     test_cases = [
@@ -11,7 +11,7 @@ def test_map_to_aro():
         ["(Phe)cpt_strepv:U09991:AAB36569:1412-1948:537", "argannot"]
     ]
 
-    ARO = pronto.Ontology.from_obo_library('aro.obo')
+    ARO = lib.get_aro_ontology()
     expected_output = [
         ARO.get_term('ARO:3002563'),
         ARO.get_term('ARO:3004623'),