Merge pull request #104 from andrewjpage/print_excluded_sequences

Support for bipartition trees as input

VERSION

@@ -1 +1 @@
-0.1.10
+0.1.11

debian/changelog

@@ -1,3 +1,9 @@
+gubbins (0.1.11~trusty1) trusty; urgency=low
+
+  * Python remove internal node names if passed in as a starting tree
+
+ -- Andrew Page <[email protected]>  Fri, 22 Aug 2014 16:44:00 +0000
+
 gubbins (0.1.10~trusty1) trusty; urgency=low
 
   * Python print list of excluded sequences to STDOUT by default

python/gubbins/common.py

@@ -634,6 +634,7 @@ def filter_out_removed_taxa_from_tree_and_return_new_file(starting_tree, temp_wo
 
     tree  = dendropy.Tree.get_from_path(starting_tree, 'newick',
               preserve_underscores=True)
+
     tree.prune_taxa_with_labels(taxa_removed, update_splits=True, delete_outdegree_one=False)          
     tree.prune_leaves_without_taxa(update_splits=True, delete_outdegree_one=False)
     tree.deroot()
@@ -642,8 +643,8 @@ def filter_out_removed_taxa_from_tree_and_return_new_file(starting_tree, temp_wo
       taxon_set=None,
       suppress_leaf_taxon_labels=False,
       suppress_leaf_node_labels=True,
-      suppress_internal_taxon_labels=False,
-      suppress_internal_node_labels=False,
+      suppress_internal_taxon_labels=True,
+      suppress_internal_node_labels=True,
       suppress_rooting=True,
       suppress_edge_lengths=False,
       unquoted_underscores=True,

python/gubbins/tests/data/robinson_foulds_distance_tree1.tre.filter_out_removed_taxa_from_tree_expected

@@ -1 +1 @@
-(((sequence_7:0.300658,sequence_10:2e-05)N3:1.1e-05,sequence_9:0.193531)N2:2e-05,sequence_6:0.030557,sequence_8:2e-05)N1:0.0;
+(((sequence_7:0.300658,sequence_10:2e-05):1.1e-05,sequence_9:0.193531):2e-05,sequence_6:0.030557,sequence_8:2e-05):0.0;

python/gubbins/tests/data/tree_with_internal_nodes.tre

@@ -0,0 +1 @@
+((A0006:0.000000,A0004:21.489000)N2:103.770966,A0003:0.000000,(A0005:0.000000,(A0001:0.000000,A0002:3.659100)N4:2.366730)N3:19.463560)N1:0.000000;

python/gubbins/tests/data/tree_with_internal_nodes.tre_expected

@@ -0,0 +1 @@
+((A0006:0.0,A0004:21.489):103.770966,A0003:0.0,(A0005:0.0,(A0001:0.0,A0002:3.6591):2.36673):19.46356):0.0;

python/gubbins/tests/test_tree_python_methods.py

@@ -76,6 +76,16 @@ def test_filter_out_removed_taxa_from_tree_and_return_new_file(self):
     os.remove(temp_working_dir + '/robinson_foulds_distance_tree1.tre')
     os.removedirs(temp_working_dir)
 
+  def test_internal_node_taxons_removed_when_used_as_starting_tree(self):
+    temp_working_dir = tempfile.mkdtemp(dir=os.getcwd())
+    common.GubbinsCommon.filter_out_removed_taxa_from_tree_and_return_new_file('gubbins/tests/data/tree_with_internal_nodes.tre', temp_working_dir, [])    
+    actual_file_content = open(temp_working_dir + '/tree_with_internal_nodes.tre', 'U').readlines()
+    expected_file_content = open('gubbins/tests/data/tree_with_internal_nodes.tre_expected', 'U').readlines()
+    assert actual_file_content == expected_file_content
+    os.remove(temp_working_dir + '/tree_with_internal_nodes.tre')
+    os.removedirs(temp_working_dir)
+
+
   def test_create_pairwise_newick_tree(self):
     common.GubbinsCommon.create_pairwise_newick_tree(['sequence_2','sequence_3'], 'gubbins/tests/data/pairwise_newick_tree.actual')
     actual_file_content = open('gubbins/tests/data/pairwise_newick_tree.actual', 'U').readlines()

python/setup.py

@@ -4,7 +4,7 @@
 
 setup(
     name='gubbins',
-    version='0.1.10',
+    version='0.1.11',
     description='Frontend to the Gubbins BioInformatics tool',
     author='Andrew J. Page',
     author_email='[email protected]',