Fixes a bug when revcomp was leaving ambiguous bases unchanged

pyfastaq/sequences.py

@@ -151,7 +151,9 @@ class Fasta:
     '''Class to store and manipulate FASTA sequences. They have two things: a name and a sequence'''
     # this defines the line length when printing sequences
     line_length = 60
-
+    # cached translation object for reverse-complementation (with ambiguous codes)
+    _revcomp_trans = str.maketrans("ATCGatcgMRWSYKVHDBNmrwsykvhdbn", 
+            "TAGCtagcKYWSRMBDHVNkywsrmbdhvn")
     def _get_id_from_header_line(self, line):
         if line.startswith('>'):
             return line.rstrip()[1:]
@@ -214,7 +216,7 @@ def strip_illumina_suffix(self):
 
     def revcomp(self):
         '''Reverse complements the sequence'''
-        self.seq = self.seq.translate(str.maketrans("ATCGatcg", "TAGCtagc"))[::-1]
+        self.seq = self.seq.translate(self._revcomp_trans)[::-1]
 
     def is_all_Ns(self, start=0, end=None):
         '''Returns true if the sequence is all Ns (upper or lower case)'''

pyfastaq/tests/sequences_test.py

@@ -125,6 +125,12 @@ def test_revcomp(self):
         fa.revcomp()
         self.assertEqual(fa, sequences.Fasta('ID', 'nacgtNACGT'))
 
+    def test_revcomp_ambig(self):
+        '''revcomp() should correctly reverse complement a sequence with ambiguous bases'''
+        fa = sequences.Fasta('ID', 'ACTGMRWSYKVHDBNmrwsykvhdbn')
+        fa.revcomp()
+        self.assertEqual(fa, sequences.Fasta('ID', 'nvhdbmrswykNVHDBMRSWYKCAGT'))
+
     def test_gaps(self):
         '''gaps() should find the gaps in a sequence correctly'''
         test_seqs = [sequences.Fasta('ID', 'ACGT'),