Initial commit

.gitignore

@@ -0,0 +1,4 @@
+__pycache__/
+*.egg-info/
+.pytest_cache/
+dist/

pyproject.toml

@@ -0,0 +1,3 @@
+[build-system]
+build-backend = 'setuptools.build_meta'
+requires = ['setuptools >= 43.0.0']

setup.cfg

@@ -0,0 +1,21 @@
+[metadata]
+name = pango_aliasor
+version = 0.0.1
+author = Cornelius Roemer
+author_email = [email protected]
+description = Pango lineage aliasing and dealiasing
+url = https://github.com/corneliusroemer/pango_aliasor
+# project_urls =
+#     Bug Tracker = https://github.com/pypa/sampleproject/issues
+classifiers =
+    Programming Language :: Python :: 3
+    Operating System :: OS Independent
+
+[options]
+package_dir = 
+    = src
+packages = find:
+python_requires = >=3.7
+
+[options.packages.find]
+where = src

src/pango_aliasor/aliasor.py

@@ -0,0 +1,46 @@
+#%%
+class Aliasor:
+    def __init__(self, alias_file=None):
+        import pandas as pd
+
+        if alias_file is None:
+            import importlib.resources
+
+            with importlib.resources.open_text("pango_designation", "alias_key.json") as file:
+                aliases = pd.read_json(file)
+
+        else:
+            aliases = pd.read_json(alias_file)
+
+        self.alias_dict = {}
+        for column in aliases.columns:
+            if column.startswith('X'):
+                self.alias_dict[column] = column
+            else:
+                self.alias_dict[column] = aliases[column][0]
+
+        self.alias_dict['A'] = 'A'
+        self.alias_dict['B'] = 'B'
+
+        self.realias_dict = {v: k for k, v in self.alias_dict.items()}
+
+    def compress(self,name):
+        name_split = name.split('.')
+        if len(name_split) < 5:
+            return name
+        letter = self.realias_dict[".".join(name_split[0:4])]
+        if len(name_split) == 5:
+            return letter + '.' + name_split[4]
+        else:
+            return letter + '.' + ".".join(name_split[4:])
+
+    def uncompress(self,name):
+        name_split = name.split('.')
+        letter = name_split[0]
+        unaliased = self.alias_dict[letter]
+        if len(name_split) == 1:
+            return name
+        if len(name_split) == 2:
+            return unaliased + '.' + name_split[1]
+        else:
+            return unaliased + '.' + ".".join(name_split[1:])

tests/test_aliasor.py

@@ -0,0 +1,24 @@
+from pango_aliasor.aliasor import Aliasor
+
+def test_pango_designation_init():
+    aliasor = Aliasor()
+
+def test_decompression():
+    aliasor = Aliasor()
+    assert aliasor.uncompress('BA.1') == 'B.1.1.529.1'
+    assert aliasor.uncompress('AY.4') == 'B.1.617.2.4'
+    assert aliasor.uncompress('AY.4.3.2') == 'B.1.617.2.4.3.2'
+    assert aliasor.uncompress('B.1') == 'B.1'
+
+def test_compression():
+    aliasor = Aliasor()
+    assert aliasor.compress('B.1.1.529.1') == 'BA.1'
+    assert aliasor.compress('B.1.617.2.4') == 'AY.4'
+    assert aliasor.compress('B.1.617.2.4.3.1') == 'AY.4.3.1'
+    assert aliasor.compress('B.1.617.2') == 'B.1.617.2'
+    assert aliasor.compress('B.1') == 'B.1'
+
+def test_except_recombinants():
+    aliasor = Aliasor()
+    assert aliasor.uncompress('XA.1') == 'XA.1'
+    assert aliasor.compress('XA.1') == 'XA.1'