🚧 Error when translating sequence lengths indivisible by 3

… instead of silently padding with N to translate to 'X'. A sequence
with 1 or 2 extra bases indicates a problem with the data. It doesn't
make sense to end the translation with 'X' which implies 3 bases.

🚧: see FIXME in code

Author: victorlin

augur/translate.py

@@ -51,25 +51,26 @@ def safe_translate(sequence):
     >>> safe_translate("")
     ''
     >>> safe_translate("ATGT")
-    'MX'
+    Traceback (most recent call last):
+    ...
+    ValueError: Sequence not divisible by 3.
     """
     from Bio.Data.CodonTable import TranslationError
     from Bio.Seq import CodonTable
 
-    #sequences not mod 3 give messy BiopythonWarning, so avoid by padding.
     if len(sequence)%3:
-        sequence_padded = sequence + "N"*(3-len(sequence)%3)
-    else:
-        sequence_padded = sequence
+        raise ValueError("Sequence not divisible by 3.")
+        # FIXME: handle this elsewhere since a proper error message needs more context - which sequence from which file?
+
     try:
         # Attempt translation by extracting the sequence according to the
         # BioPhython SeqFeature in frame gaps of three will translate as '-'
-        translated_sequence = str(Seq.Seq(sequence_padded).translate(gap='-'))
+        translated_sequence = str(Seq.Seq(sequence).translate(gap='-'))
     except TranslationError:
         # Any other codon like '-AA' or 'NNT' etc will fail. Translate codons
         # one by one.
         codon_table  = CodonTable.ambiguous_dna_by_name['Standard'].forward_table
-        str_seq = str(sequence_padded)
+        str_seq = str(sequence)
         codons = np.frombuffer(str_seq[:len(str_seq) - len(str_seq) % 3].encode(), dtype='S3').astype("U")
         assert len(codons) > 0
         aas = []

tests/test_translate.py

@@ -4,6 +4,7 @@
 from pathlib import Path
 import sys
 
+import pytest
 from Bio.Seq import Seq
 from Bio.SeqFeature import SeqFeature, FeatureLocation
 
@@ -23,14 +24,29 @@ def test_safe_translate(self):
                            (('ATG---',), 'M-'),
                            (('ATGTAG',), 'M*'),
                            (('',), ''),
-                           (('ATGT',), 'MX'),
                            (('ATGA-G',), 'MX')]
 
         # input each pair into the function and check
         for pair in params_and_outs:
             params, out = pair
             assert translate.safe_translate(*params) == out
 
+    def test_safe_translate_errors(self):
+        '''
+        Test that safe_translate raises ValueError when sequence is not divisible by 3
+        '''
+        invalid_sequences = [
+            'A',
+            'AT',
+            'ATGT',
+            'ATGTA',
+            'ATGTAGA',
+        ]
+
+        for seq in invalid_sequences:
+            with pytest.raises(ValueError, match="Sequence not divisible by 3"):
+                translate.safe_translate(seq)
+
     def test_translate_feature(self):
         '''
         Test translate_feature from a dictionary of given nucleotides to dictionary of translated amino acids