Add the convert_to_itol function for #373

gtdbtk/__main__.py

@@ -48,11 +48,13 @@ 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
-    remove_labels -> Remove labels (bootstrap values, node labels) from an Newick tree
+    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
+    convert_to_itol -> Convert a GTDB-Tk Newick tree to an iTOL tree
+ 
 
   Testing:
     test          -> Validate the classify_wf pipeline with 3 archaeal genomes 

gtdbtk/cli.py

@@ -535,7 +535,16 @@ def get_main_parser():
 
     # 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:
+                                                 '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)
+
+    # Remove labels
+    with subparser(sub_parsers, 'convert_to_itol', 'Reformat the GTDB-Tk tree to be iTOL compatible.') as parser:
         with arg_group(parser, 'required named arguments') as grp:
             __input_tree(grp, required=True)
             __output_tree(grp, required=True)

gtdbtk/config/config.py

@@ -320,7 +320,7 @@
               MASK_DIR: '50e414a9de18170e8cb97f990f89ff60a0fe29d5',
               MARKER_DIR: '163f542c3f0a40f59df45d453aa235b39aa96e27',
               RADII_DIR: '8fd13b1c5d7a7b073ba96fb628581613b293a374',
-              MSA_FOLDER: '4bd032c90d5e5f0cbc96338445721a317f7d90b4',
+              MSA_FOLDER: '24f250d7cf0eb0bc65dccd2f3c9247e553ea322f',
               METADATA_DIR: '9772fbeac1311b31e10293fa610eb33aa1ec8e15',
               TAX_FOLDER: '6fb0233b05633242369b40c026fd1ee53e266afa',
               FASTANI_DIR: '973c456c02f55bb82908a6811c7076e207e9b206',

gtdbtk/main.py

@@ -396,7 +396,7 @@ def run_test(self, options):
             shutil.copytree(input_dir, genome_test_dir)
 
             args = ['gtdbtk', 'classify_wf', '--genome_dir', genome_test_dir,
-                    '--out_dir', output_dir, '--cpus', str(options.cpus)]
+                    '--out_dir', output_dir, '--cpus', str(options.cpus), '-f']
             self.logger.info('Command: {}'.format(' '.join(args)))
 
             # Pipe the output and write to disk.
@@ -464,6 +464,10 @@ def classify(self, options):
                      fulltreeopt=options.full_tree,
                      recalculate_red=False)
 
+        self.logger.info('Note that Tk classification mode is insufficient for publication of new taxonomic '
+                         'designations. New designations should be based on one or more de novo trees, an '
+                         'example of which can be produced by Tk in de novo mode.')
+
         self.logger.info('Done.')
 
     def trim_msa(self, options):
@@ -624,6 +628,20 @@ def ani_rep(self, options):
 
         self.logger.info('Done.')
 
+    def convert_to_itol(self, options):
+        """Convert Tree to iTOL format.
+
+        Parameters
+        ----------
+        options : argparse.Namespace
+            The CLI arguments input by the user.
+        """
+        check_file_exists(options.input_tree)
+
+        r = Misc()
+        r.convert_to_itol(options.input_tree, options.output_tree)
+        self.logger.info('Done.')
+
     def remove_labels(self, options):
         """Remove labels from tree.
 
@@ -826,6 +844,8 @@ def parse_options(self, options):
             self.ani_rep(options)
         elif options.subparser_name == 'remove_labels':
             self.remove_labels(options)
+        elif options.subparser_name == 'convert_to_itol':
+            self.convert_to_itol(options)
         elif options.subparser_name == 'trim_msa':
             self.trim_msa(options)
         elif options.subparser_name == 'export_msa':

gtdbtk/misc.py

@@ -25,6 +25,7 @@
 import gtdbtk.config.config as Config
 from gtdbtk.biolib_lite.execute import check_dependencies
 from gtdbtk.biolib_lite.logger import colour
+from gtdbtk.biolib_lite.newick import parse_label
 from gtdbtk.biolib_lite.seq_io import read_fasta
 from gtdbtk.config.output import DIR_CLASSIFY_INTERMEDIATE, DIR_ALIGN_INTERMEDIATE, DIR_IDENTIFY_INTERMEDIATE
 from gtdbtk.exceptions import GTDBTkException, GTDBTkExit
@@ -126,7 +127,36 @@ def remove_labels(self, input_file, output_file):
         for node in intree.internal_nodes():
             node.label = None
 
-        intree.write_to_path(output_file, schema='newick', suppress_rooting=True)
+        intree.write_to_path(output_file, schema='newick', suppress_rooting=True,unquoted_underscores=True)
+
+
+    def convert_to_itol(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("Convert GTDB-Tk tree to iTOL format")
+        intree= dendropy.Tree.get_from_path(input_file,
+                                           schema='newick',
+                                           rooting='force-rooted',
+                                           preserve_underscores=True)
+
+        for node in intree.internal_nodes():
+            if node.label:
+                bootstrap,label,_aux = parse_label(node.label)
+                if label:
+                    label = label.replace('; ',';').replace(';','|').replace("'","").lstrip('')
+                node.label = label
+                if node.edge.length:
+                    node.edge.length = f'{node.edge.length}[{bootstrap}]'
+
+        intree.write_to_path(output_file, schema='newick', suppress_rooting=True,unquoted_underscores=True)
 
 
     def remove_intermediate_files(self,output_dir,wf_name):