GenBank: Don't discard topology if given in LOCUS line

Rebased from pull request biopython#812 to address biopython#363, with
minor changes as discussed on the pull request.

Bio/GenBank/__init__.py

@@ -1241,6 +1241,12 @@ def record_end(self, content):
                 raise ValueError("Could not determine alphabet for seq_type %s"
                                  % self._seq_type)
 
+            # Also save the chomosome layout
+            if 'circular' in self._seq_type.lower():
+                self.data.annotations['topology'] = 'circular'
+            elif 'linear' in self._seq_type.lower():
+                self.data.annotations['topology'] = 'linear'
+
         if not sequence and self.__expected_size:
             self.data.seq = UnknownSeq(self._expected_size, seq_alphabet)
         else:

Bio/SeqIO/InsdcIO.py

@@ -557,6 +557,18 @@ def _get_data_division(self, record):
         assert len(division) == 3
         return division
 
+    def _get_topology(self, record):
+        """Set the topology to 'circular', 'linear' if defined"""
+        max_topology_len = len('circular')
+
+        # return an empty placeholder string if not given
+        if 'topology' not in record.annotations:
+            return ' ' * max_topology_len
+
+        template = '%%-%ds' % max_topology_len
+
+        return template % record.annotations['topology'][:max_topology_len]
+
     def _write_the_first_line(self, record):
         """Write the LOCUS line."""
 
@@ -610,6 +622,8 @@ def _write_the_first_line(self, record):
             # just the generic Alphabet (default for fasta files)
             raise ValueError("Need a DNA, RNA or Protein alphabet")
 
+        topology = self._get_topology(record)
+
         division = self._get_data_division(record)
 
         name_length = str(len(record)).rjust(28)
@@ -619,12 +633,11 @@ def _write_the_first_line(self, record):
 
         assert len(units) == 2
         assert len(division) == 3
-        # TODO - date
-        # TODO - mol_type
-        line = "LOCUS       %s %s    %s           %s %s\n" \
+        line = "LOCUS       %s %s    %s %s %s %s\n" \
             % (name_length,
                units,
-               mol_type.ljust(6),
+               mol_type.ljust(7),
+               topology,
                division,
                self._get_date(record))
         assert len(line) == 79 + 1, repr(line)  # plus one for new line

Tests/test_GenBank_unittest.py

@@ -161,6 +161,18 @@ def test_structured_comment_parsing(self):
                          'reference sequence was derived from AP000423.\n'
                          'COMPLETENESS: full length.')
 
+    def test_locus_line_topogoly(self):
+        """Test if chromosome topology is conserved"""
+        record = SeqIO.read('GenBank/DS830848.gb', 'genbank')
+        self.assertEqual(record.annotations['topology'], 'linear')
+        out_handle = StringIO()
+        SeqIO.write([record], out_handle, 'genbank')
+        first_line = out_handle.getvalue().split('\n')[0]
+        self.assertIn('linear', first_line)
+        with open('GenBank/DS830848.gb', 'r') as fh:
+            orig_first_line = fh.readline().strip()
+        self.assertEqual(first_line, orig_first_line)
+
     def test_long_names(self):
         """Various GenBank names which push the column based LOCUS line."""
         original = SeqIO.read("GenBank/iro.gb", "gb")