[nomenclature._GPCRdbDataFrame2conlabs] recover GAIN labels even if scheme is not display_generic_number, tests

Author: gph82

mdciao/filenames/filenames.py

@@ -101,6 +101,7 @@ def __init__(self):
         self.adrb2_human_xlsx = _path.join(self.nomenclature_path,"adrb2_human.xlsx")
         self.nomenclature_bib = _path.join(self.nomenclature_path,"nomenclature.bib")
         self.KLIFS_P31751_xlsx = _path.join(self.nomenclature_path, "KLIFS_P31751.xlsx")
+        self.agrg1_human_xlsx = _path.join(self.nomenclature_path, "agrg1_human.xlsx")
 
         #json
         self.GDP_json = _path.join(self.json_path,"GDP.json")

mdciao/nomenclature/nomenclature.py

@@ -106,6 +106,14 @@ def _GPCRdbDataFrame2conlabs(tablefile,
     assert scheme in df.keys(), ValueError("'%s' isn't an available scheme.\nAvailable schemes are %s" % (
         scheme, [key for key in df.keys() if key in _GPCR_available_schemes + ["display_generic_number"]]))
     AA2conlab = {key: str(val) for key, val in df[["AAresSeq", scheme]].values}
+    dgn_other = df[df[scheme].isna() & ~df["display_generic_number"].isna()][["AAresSeq", "display_generic_number"]]
+    if len(dgn_other)>0:
+        AA2dgn = {key: str(val) for key, val in df[["AAresSeq", "display_generic_number"]].values}
+        for ii, key in enumerate(AA2conlab.keys()):
+            if str(AA2conlab[key]) == "None" and AA2dgn.get(key,None) is not None:
+                AA2conlab[key]=AA2dgn[key]
+        #print(f"Some labels of the 'display_generic_number' column would be lost by choosing scheme='{scheme}'.")
+
     # Locate definition lines and use their indices
     fragments = _defdict(list)
 
@@ -1246,7 +1254,7 @@ def __init__(self, UniProt_name,
             including the extension "xslx", then the lookup will
             fail. This what the `format` parameter is for
         write_to_disk : bool, default is False
-            Save an excel file with the nomenclature
+            Save an Excel file with the nomenclature
             information
         """
 
@@ -1258,7 +1266,8 @@ def __init__(self, UniProt_name,
                                                           write_to_disk=write_to_disk
                                                           )
         # Re-introduce the "." in the GPS label
-        self._dataframe = self._dataframe.replace("B.GPS-2","B.GPS.-2").replace("B.GPS-1","B.GPS.-1").replace("B.GPS+1","B.GPS.+1")
+        self._dataframe = self._dataframe.replace("B.GPS-2", "B.GPS.-2").replace("B.GPS-1", "B.GPS.-1").replace("B.GPS+1", "B.GPS.+1")
+
         # Check for GPS in the middle of S14
         if "B.S14" in self.dataframe.protein_segment.values:
             assert self.dataframe[self.dataframe.protein_segment == "B.S14"].index.diff().fillna(1).unique() == 1, (

tests/data/nomenclature/agrg1_human.xlsx

@@ -156,6 +156,29 @@ def test_wont_fail_if_found_online_and_write_to_disk(self):
         _np.testing.assert_array_equal(list(df.keys())[:3], nomenclature._GPCR_mandatory_fields)
         assert any([key in df.keys() for key in nomenclature._GPCR_mandatory_fields])  # at least one scheme
 
+class Test_GPCRdbDataFrame2conlabs_aGPCR(unittest.TestCase):
+    def setUp(self):
+        self.file = test_filenames.agrg1_human_xlsx
+
+    def test_works(self):
+        pass
+        CL = nomenclature.LabelerGPCRdb(self.file)
+
+    def test_works_scheme_BW(self):
+        CL_generic = nomenclature.LabelerGPCRdb(self.file)
+        CL_BW = nomenclature.LabelerGPCRdb(self.file, scheme="BW")
+        self.assertDictEqual(CL_generic.fragments, CL_BW.fragments)
+
+        #self.assertDictEqual(CL_generic.AA2conlab, CL_BW.AA2conlab)
+        # This assertion will fail because some residues of the TM bundle have different labels
+        # on display_generic_number and BW, e.g 1.32x32 vs 1.28 on D396
+        # protein_segment AAresSeq display_generic_number    BW Wootten   Pin  Wang Fungal GPCRdb(A) GPCRdb(B) GPCRdb(C) GPCRdb(F) GPCRdb(D) Oliveira     BS
+        #             TM1     D396                1.32x32  1.28    1.32  1.32  1.25   1.30   1.28x28   1.32x32   1.32x32   1.25x25   1.30x30      108  I:-05
+        # We thus only test the GAIN domain
+        generic_GAIN_conlab = {key : val for key, val in CL_generic.AA2conlab.items() if key[0] in ["A","B"]}
+        BW_GAIN_conlab = {key: val for key, val in CL_generic.AA2conlab.items() if key[0] in ["A", "B"]}
+        self.assertDictEqual(generic_GAIN_conlab, BW_GAIN_conlab)
+
 
 class Test_GPCRdbDataFrame2conlabs(unittest.TestCase):
     def setUp(self):