Use OrderedDict for SeqFeature qualifiers.

Part of this behaviour will happen by default as of Python 3.6 where the
Python dict preserves order, but goal here is to help round-trip EMBL/GenBank
files by preserving the feature qualifier ordering.

This means we can stop explicitly sorting the qualifiers on output (which
produced stable but arbitrary ordering, rather than attempting to match the
original order).

Bio/SeqFeature.py

@@ -1,6 +1,6 @@
 # Copyright 2000-2003 Jeff Chang.
 # Copyright 2001-2008 Brad Chapman.
-# Copyright 2005-2015 by Peter Cock.
+# Copyright 2005-2016 by Peter Cock.
 # Copyright 2006-2009 Michiel de Hoon.
 # All rights reserved.
 # This code is part of the Biopython distribution and governed by its
@@ -54,6 +54,8 @@
 
 from __future__ import print_function
 
+from collections import OrderedDict
+
 from Bio._py3k import _is_int_or_long
 
 from Bio.Seq import MutableSeq, reverse_complement
@@ -85,7 +87,7 @@ class SeqFeature(object):
         - qualifiers - A dictionary of qualifiers on the feature. These are
           analogous to the qualifiers from a GenBank feature table. The keys of
           the dictionary are qualifier names, the values are the qualifier
-          values.
+          values. As of Biopython 1.69 this is an ordered dictionary.
     """
 
     def __init__(self, location=None, type='', location_operator='',
@@ -150,7 +152,7 @@ def __init__(self, location=None, type='', location_operator='',
             self.strand = strand
         self.id = id
         if qualifiers is None:
-            qualifiers = {}
+            qualifiers = OrderedDict()
         self.qualifiers = qualifiers
         if sub_features is not None:
             raise TypeError("Rather than sub_features, use a CompoundFeatureLocation")
@@ -271,7 +273,7 @@ def _shift(self, offset):
                           type=self.type,
                           location_operator=self.location_operator,
                           id=self.id,
-                          qualifiers=dict(self.qualifiers.items()))
+                          qualifiers=OrderedDict(self.qualifiers.items()))
 
     def _flip(self, length):
         """Returns a copy of the feature with its location flipped (PRIVATE).
@@ -287,7 +289,7 @@ def _flip(self, length):
                           type=self.type,
                           location_operator=self.location_operator,
                           id=self.id,
-                          qualifiers=dict(self.qualifiers.items()))
+                          qualifiers=OrderedDict(self.qualifiers.items()))
 
     def extract(self, parent_sequence):
         """Extract feature sequence from the supplied parent sequence.

Bio/SeqIO/InsdcIO.py

@@ -357,8 +357,8 @@ def _write_feature(self, feature, record_length):
                + self._wrap_location(location) + "\n"
         self.handle.write(line)
         # Now the qualifiers...
-        for key in sorted(feature.qualifiers.keys()):
-            values = feature.qualifiers[key]
+        # Note as of Biopython 1.69, this is an ordered-dict, don't sort it:
+        for key, values in feature.qualifiers.items():
             if isinstance(values, (list, tuple)):
                 for value in values:
                     self._write_feature_qualifier(key, value)

NEWS

@@ -23,6 +23,10 @@ now also allowed in identifiers.
 For consistency the Bio.Seq module now offers a complement function (already
 available as a method on the Seq and MutableSeq objects).
 
+The SeqFeature object's qualifiers is now an explicitly ordered dictionary
+(note that as of Python 3.6 the Python dict is ordered by default anyway).
+This helps reproduce GenBank/EMBL files on input/output.
+
 Additionally, a number of small bugs have been fixed with further additions
 to the test suite, and there has been further work to follow the Python PEP8,
 PEP257 and best practice standard coding style.

Tests/test_GenBank_unittest.py

@@ -175,6 +175,13 @@ def test_locus_line_topogoly(self):
             orig_first_line = fh.readline().strip()
         self.assertEqual(first_line, orig_first_line)
 
+    def test_qualifier_order(self):
+        """Check the qualifier order is preserved."""
+        record = SeqIO.read("GenBank/DS830848.gb", "gb")
+        f = record.features[0]
+        self.assertEqual(list(f.qualifiers),
+                         ['organism', 'mol_type', 'strain', 'db_xref', 'dev_stage'])
+
     def test_long_names(self):
         """Various GenBank names which push the column based LOCUS line."""
         original = SeqIO.read("GenBank/iro.gb", "gb")