Classyfire updates (#92)

* deleting old test data

* url encode smiles as there may be reserved characters

* add an example smile with special characters

* test if smile with spl. characters is classified

* check taxonomy keys in classyfire response - produces key error otherwise

* test if molecule was assigned to the correct kingdom

Author: anupriyatripathi

q2_qemistree/_classyfire.py

@@ -10,6 +10,7 @@
 import pandas as pd
 import numpy as np
 import warnings
+import urllib
 
 
 def get_classyfire_taxonomy(feature_data: pd.DataFrame) -> pd.DataFrame:
@@ -65,21 +66,26 @@ def get_classyfire_taxonomy(feature_data: pd.DataFrame) -> pd.DataFrame:
     for idx in feature_data.index:
         smiles = feature_data.loc[idx, 'smiles']
         if pd.notna(smiles):
-            url_smiles = 'https://gnps-structure.ucsd.edu/inchikey?smiles='
-            response = requests.get(url_smiles+smiles)
+            to_inchikey = 'https://gnps-structure.ucsd.edu/inchikey?smiles='
+            urlencoded_smiles = urllib.parse.quote(smiles)
+            response = requests.get(to_inchikey+urlencoded_smiles)
             if response.status_code != 200:
                 classyfire[idx] = 'SMILE parse error'
                 no_inchikey.append((idx, smiles))
                 continue
             inchikey = response.text
-            url_inchi = 'https://gnps-classyfire.ucsd.edu/entities/'
-            response = requests.get(url_inchi+str(inchikey)+'.json')
+            to_classyfire = 'https://gnps-classyfire.ucsd.edu/entities/'
+            response = requests.get(to_classyfire+str(inchikey)+'.json')
             if response.status_code == 200:
                 response = response.json()
+                classyfire_levels_sub = [level for level in classyfire_levels
+                                         if level in response.keys()]
+                if len(classyfire_levels_sub) == 0:
+                    continue
                 taxonomy = [response[level]['name']
                             if bool(response) and response[level] is not None
                             else 'unclassified'
-                            for level in classyfire_levels]
+                            for level in classyfire_levels_sub]
                 classyfire[idx] = taxonomy
             elif response.status_code == 404:
                 classyfire[idx] = 'unclassified'

q2_qemistree/tests/data/feature_data_smiles.txt

@@ -1,4 +1,4 @@
 label	#featureID	csi_smiles	ms2_smiles	table_number
 3638d9b9658b202f9c773e5e43206d32	3			1
-c8f6e296c0bacf58cfec06b333fd2015	7	CC(=N)C1=CN(C)C=N1		1
+c8f6e296c0bacf58cfec06b333fd2015	7	CCCCCCCCCCCCCCCC(=O)OCC(COP(=O)([O-])OCC[N+](C)(C)C)O		1
 a500d26568f130313bdb70e3ae7d2ade	2			1

q2_qemistree/tests/test_classyfire.py

@@ -52,9 +52,12 @@ def test_malformed_smiles(self):
             get_classyfire_taxonomy(self.mal_smiles)
 
     def test_classyfire_output(self):
-        classified_feature_data = get_classyfire_taxonomy(self.smiles)
-        self.assertTrue((self.levels.issubset(set(
-            classified_feature_data.columns))))
+        classified = get_classyfire_taxonomy(self.smiles)
+        classified_mols = classified[classified['kingdom'] != 'unclassified']
+        self.assertTrue(len(classified_mols) == 1)
+        self.assertTrue(classified_mols.loc[1,
+                        'kingdom'] == 'Organic compounds')
+        self.assertTrue((self.levels.issubset(set(classified.columns))))
 
 
 if __name__ == '__main__':