no need to check if current gene_code is different

voseq/create_dataset/tests/tests_utils.py

@@ -28,9 +28,10 @@ def setUp(self):
 
         self.c = Client()
         self.dataset_creator = CreateDataset(self.cleaned_data)
+        self.maxDiff = None
 
     def test_create_dataset(self):
-        expected = '>CP100-10_Papilionoidea_Melitaea_diamina'
+        expected = '>coi\n--------------------\n>CP100-10_Papilionoidea_Melitaea_diamina'
         result = self.dataset_creator.dataset_str
         self.assertTrue(expected in result)
 
@@ -47,9 +48,13 @@ def test_get_taxon_names_for_taxa(self):
             'cp100-11': {'code': 'CP100-11', 'genus': 'Melitaea', 'species': 'diamina', 'superfamily': ''},
         }
         result = self.dataset_creator.get_taxon_names_for_taxa()
-
         self.assertEqual(expected, result)
 
+    def test_from_seq_objs_to_fasta(self):
+        expected = 2706
+        result = self.dataset_creator.from_seq_objs_to_fasta()
+        self.assertEqual(expected, len(result))
+
     def test_get_taxon_names_for_taxa_additional_fields(self):
         self.cleaned_data['taxon_names'] = ['SUPERFAMILY']
         dataset_creator = CreateDataset(self.cleaned_data)

voseq/create_dataset/utils.py

@@ -82,10 +82,6 @@ def from_seq_objs_to_fasta(self):
                     this_gene = seq_record.name
                     seq_str = '>' + this_gene + '\n' + '--------------------'
                     append(seq_str)
-                if this_gene != seq_record.name:
-                    this_gene = seq_record.name
-                    seq_str = '>' + this_gene + '\n' + '--------------------'
-                    append(seq_str)
                 seq_record_seq_str = str(self.get_sequence_based_on_codon_positions(this_gene, seq_record.seq))
                 seq_str = '>' + seq_record.id + '\n' + seq_record_seq_str
                 append(seq_str)