ete3: allow taxids in species tree input

now one can submit a list of taxids as input
(or even mixed)

Author: bernt-matthias

tools/ete/ete_species_tree_generator.py

@@ -6,7 +6,7 @@
 
 parser = optparse.OptionParser()
 parser.add_option('-s', '--species', dest="input_species_filename",
-                  help='Species list in text format one species in each line')
+                  help='List of species names of taxids in text format one species in each line')
 parser.add_option('-d', '--database', dest="database", default=None,
                   help='ETE sqlite data base to use (default: ~/.etetoolkit/taxa.sqlite)')
 parser.add_option('-o', '--output', dest="output", help='output file name (default: stdout)')
@@ -19,26 +19,32 @@
     parser.error("-s option must be specified, Species list in text format one species in each line")
 
 ncbi = NCBITaxa(dbfile=options.database)
-with open(options.input_species_filename) as f:
-    species_name = [_.strip().replace('_', ' ') for _ in f.readlines()]
-
-name2taxid = ncbi.get_name_translator(species_name)
 
-taxid = [name2taxid[_][0] for _ in species_name]
-
-tree = ncbi.get_topology(taxid)
+# determine taxids and species names in the input file
+names = []
+taxids = []
+with open(options.input_species_filename) as f:
+    for species in f:
+        species = species.strip().replace('_', ' ')
+        try:
+            taxids.append(int(species))
+        except ValueError:
+            names.append(species)
+# translate all species names to taxids
+name2taxid = ncbi.get_name_translator(names)
+taxids += {name2taxid[n][0]  for n in names}
+
+# get topology and set the scientific name as output
+tree = ncbi.get_topology(taxids)
+for isleaf, node in tree.iter_prepostorder():
+    node.name = node.sci_name
 
 if options.treebest == "yes":
-    inv_map = {str(v[0]): k.replace(" ", "") + "*" for k, v in name2taxid.items()}
-else:
-    inv_map = {str(v[0]): k for k, v in name2taxid.items()}
-
-
-for leaf in tree:
-    leaf.name = inv_map[leaf.name]
+    for leaf in tree:
+        leaf.name = leaf.name.replace(" ", "") + "*"
 
 newickTree = tree.write(format=int(options.format))
-
+# print(type(tree))
 if options.treebest == "yes":
     newickTree = newickTree.rstrip(';')
     newickTree = newickTree + "root;"

tools/ete/ete_species_tree_generator.xml

@@ -1,4 +1,4 @@
-<tool id="ete_species_tree_generator" name="ETE species tree generator" version="@VERSION@">
+<tool id="ete_species_tree_generator" name="ETE species tree generator" version="@VERSION@+galaxy1">
     <description>from a list of species using the ETE Toolkit</description>
     <macros>
         <import>ete_macros.xml</import>
@@ -21,7 +21,7 @@ python '$__tool_directory__/ete_species_tree_generator.py'
 -t $output_format.treebest
     ]]></command>
     <inputs>
-        <param name="speciesFile" type="data" format="txt" label="Species file" help="List with one species per line" />
+        <param name="speciesFile" type="data" format="txt" label="Species file" help="List with one species name or taxid per line" />
         <param name="database" type="data" format="sqlite" label="(ETE3) Taxonomy Database" help="The sqlite formatted Taxonomy used by ETE3 (which is derived from NCBI taxonomy)" />
         <conditional name="output_format">
             <param name="treebest" type="select" label="Use in TreeBest" help="Select yes if specie tree to be used in TreeBest">

tools/ete/test-data/species.txt

@@ -28,7 +28,7 @@ Nomascus leucogenys
 Pongo abelii
 Gorilla gorilla gorilla
 Pan troglodytes
-Homo sapiens
+9606
 Sorex araneus
 Erinaceus europaeus
 Pteropus vampyrus