Add the remove_labels function for #373, removing internal labels mak…

…es the tree compatible with Itol

docs/src/installing/index.rst

@@ -35,11 +35,11 @@ Hardware requirements
      - Time
    * - Archaea
      - ~34 GB
-     - ~30 GB
+     - ~65 GB
      - ~1 hour / 1,000 genomes @ 64 CPUs
    * - Bacteria
      - ~320 GB ( 20GB for divide-and-conquer)
-     - ~30 GB
+     - ~65 GB
      - ~1 hour / 1,000 genomes @ 64 CPUs
 
 .. note::

gtdbtk/__main__.py

@@ -48,10 +48,11 @@ def print_help():
     decorate -> Decorate tree with GTDB taxonomy
 
   Tools:
-    infer_ranks -> Establish taxonomic ranks of internal nodes using RED
-    ani_rep     -> Calculates ANI to GTDB representative genomes
-    trim_msa    -> Trim an untrimmed MSA file based on a mask
-    export_msa  -> Export the untrimmed archaeal or bacterial MSA file
+    infer_ranks   -> Establish taxonomic ranks of internal nodes using RED
+    ani_rep       -> Calculates ANI to GTDB representative genomes
+    trim_msa      -> Trim an untrimmed MSA file based on a mask
+    export_msa    -> Export the untrimmed archaeal or bacterial MSA file
+    remove_labels -> Remove labels (bootstrap values, node labels) from an Newick tree. 
 
   Testing:
     test          -> Validate the classify_wf pipeline with 3 archaeal genomes 

gtdbtk/cli.py

@@ -524,6 +524,16 @@ def get_main_parser():
             __debug(grp)
             __help(grp)
 
+    # Remove labels
+    with subparser(sub_parsers, 'remove_labels', 'Remove labels (bootstrap values, node labels) from an Newick tree to '
+                                                 'to improve compatibility with tree viewers') as parser:
+        with arg_group(parser, 'required named arguments') as grp:
+            __input_tree(grp, required=True)
+            __output_tree(grp, required=True)
+        with arg_group(parser, 'optional arguments') as grp:
+            __debug(grp)
+            __help(grp)
+
     # Export MSA.
     with subparser(sub_parsers, 'export_msa', 'Export the untrimmed archaeal or bacterial MSA file.') as parser:
         with arg_group(parser, 'required named arguments') as grp:

gtdbtk/main.py

@@ -624,6 +624,21 @@ def ani_rep(self, options):
 
         self.logger.info('Done.')
 
+    def remove_labels(self, options):
+        """Remove labels from tree.
+
+        Parameters
+        ----------
+        options : argparse.Namespace
+            The CLI arguments input by the user.
+        """
+
+        check_file_exists(options.input_tree)
+
+        r = Misc()
+        r.remove_labels(options.input_tree, options.output_tree)
+        self.logger.info('Done.')
+
     def remove_intermediate_files(self,out_dir,workflow_name):
         """Remove intermediate files from the output directory.
         Parameters
@@ -809,6 +824,8 @@ def parse_options(self, options):
             self.infer_ranks(options)
         elif options.subparser_name == 'ani_rep':
             self.ani_rep(options)
+        elif options.subparser_name == 'remove_labels':
+            self.remove_labels(options)
         elif options.subparser_name == 'trim_msa':
             self.trim_msa(options)
         elif options.subparser_name == 'export_msa':

gtdbtk/misc.py

@@ -20,6 +20,8 @@
 
 import shutil
 
+import dendropy
+
 import gtdbtk.config.config as Config
 from gtdbtk.biolib_lite.execute import check_dependencies
 from gtdbtk.biolib_lite.logger import colour
@@ -104,6 +106,29 @@ def checkfolder(self, folder_path, folder_name):
                 folder_name, folder_path, colour('MISSING', ['bright'], fg='red')))
             return False
 
+    def remove_labels(self, input_file, output_file):
+        """Remove labels from a Newick Tree.
+
+        Parameters
+        ----------
+        input_file : str
+            The path to the input Newick tree.
+        output_file : str
+            The path to the output Newick tree.
+        """
+
+        self.logger.info("Removing labels from tree {}".format(input_file))
+        intree= dendropy.Tree.get_from_path(input_file,
+                                           schema='newick',
+                                           rooting='force-rooted',
+                                           preserve_underscores=True)
+
+        for node in intree.internal_nodes():
+            node.label = None
+
+        intree.write_to_path(output_file, schema='newick', suppress_rooting=True)
+
+
     def remove_intermediate_files(self,output_dir,wf_name):
         """Remove intermediate files.
 

tests/test_cli.py

@@ -192,6 +192,15 @@ def test_identify_align_classify(self):
         self.assertEqual(len(infos), 20)
         self.assertTrue(infos[1].startswith('d__Archaea'))
 
+        self.assertTrue(os.path.isdir(os.path.join(classify_options.out_dir, DIR_IDENTIFY_INTERMEDIATE)))
+        self.assertTrue(os.path.isdir(os.path.join(classify_options.out_dir, DIR_ALIGN_INTERMEDIATE)))
+        self.assertTrue(os.path.isdir(os.path.join(classify_options.out_dir, DIR_CLASSIFY_INTERMEDIATE)))
+        self.optionparser.remove_intermediate_files(classify_options.out_dir,'classify_wf')
+        self.assertFalse(os.path.exists(os.path.join(classify_options.out_dir, DIR_IDENTIFY_INTERMEDIATE)))
+        self.assertFalse(os.path.exists(os.path.join(classify_options.out_dir, DIR_ALIGN_INTERMEDIATE)))
+        self.assertFalse(os.path.exists(os.path.join(classify_options.out_dir, DIR_CLASSIFY_INTERMEDIATE)))
+
+
     def test_classify_wf(self):
         tmp_folder = ''.join(random.choice(
             string.ascii_uppercase + string.digits) for _ in range(10))
@@ -265,7 +274,7 @@ def test_root(self):
         """Test that rooting is successful when called through the CLI"""
         options = argparse.ArgumentParser()
         options.input_tree = 'tests/data/pplacer_dir_reference/gtdbtk.ar53.classify.tree'
-        options.outgroup_taxon = 'p__Altarchaeota'
+        options.outgroup_taxon = 'p__Altiarchaeota'
         options.output_tree = os.path.join(self.generic_out_path, 'test.rooted.tree')
         options.custom_taxonomy_file = None
         options.gtdbtk_classification_file = None