protein/orgainism counts in uniprot collection

datanator_query_python/query/query_protein.py

@@ -778,4 +778,20 @@ def get_equivalent_kegg_with_anchor(self, ko, anchor, max_distance, max_depth=fl
                     result[level]['documents'].append(equivalent)
             checked_ids.append(cur_id)
 
-        return result
+        return result
+
+    def get_unique_protein(self):
+        """Get number of unique proteins in collection
+
+        Return:
+            (:obj:`int`): number of unique proteins.
+        """
+        return len(self.collection.distinct('uniprot_id', collation=self.collation))
+
+    def get_unique_organism(self):
+        """Get number of unique organisms in collection.
+
+        Return:
+            (:obj:`int`): number of unique organisms.
+        """
+        return len(self.collection.distinct('ncbi_taxonomy_id'))

tests/query/test_query_protein.py

@@ -132,7 +132,6 @@ def test_get_info_by_text_abundances(self):
         self.assertEqual(result[0]['uniprot_ids'], {'uniprot7': False, 'uniprot0': True})
         name = 'Nucleoside diphosphate kinase'
         result = self.src_1.get_info_by_text_abundances(name)
-        print(result[0:9])
 
     def test_get_info_by_taxonid(self):
         _id = 6
@@ -254,4 +253,13 @@ def test_get_kegg_orthology(self):
 
     def test_get_equivalent_kegg_with_anchor(self):
         result_0 = self.src_1.get_equivalent_kegg_with_anchor('K03154','Thermus thermophilus HB27', 3, max_depth=2)
-        self.assertTrue(len(result_0[0]['documents']) > 0)
+        self.assertTrue(len(result_0[0]['documents']) > 0)
+
+    @unittest.skip('passed')
+    def test_get_unique_protein(self):
+        result = self.src_1.get_unique_protein()
+        print(result)
+
+    def test_get_unique_organism(self):
+        result = self.src_1.get_unique_organism()
+        print(result)