def parse_label(label):
"""Parse a Newick label which may contain a support value, taxon, and/or auxiliary information.
Parameters
----------
label : str
Internal label in a Newick tree.
Returns
-------
float
Support value specified by label, or None
str
Taxon specified by label, or None
str
Auxiliary information, on None
"""
support = None
taxon = None
auxiliary_info = None
if label:
label = label.strip()
if '|' in label:
label, auxiliary_info = label.split('|')
if ':' in label:
support, taxon = label.split(':')
support = float(support)
else:
if is_float(label):
support = float(label)
elif label != '':
taxon = label
return support, taxon, auxiliary_info
def read_from_tree(self, tree, warnings=True):
"""Obtain the taxonomy for each extant taxa as specified by internal tree labels.
Parameters
----------
tree : str or dendropy.Tree
Filename of newick tree or dendropy tree object.
Returns
-------
dict : d[unique_id] -> [d__<taxon>, ..., s__<taxon>]
Taxa indexed by unique ids.
"""
if isinstance(tree, str):
tree = dendropy.Tree.get_from_path(tree,
schema='newick',
rooting="force-rooted",
preserve_underscores=True)
taxonomy = {}
for leaf in tree.leaf_node_iter():
taxa = []
node = leaf.parent_node
while node:
if node.label:
taxa_str = node.label
if ':' in taxa_str:
taxa_str = taxa_str.split(':')[1]
if not is_float(taxa_str):
if taxa_str[-1] == ';':
taxa_str = taxa_str[:-1]
# check for concatenated ranks of the form:
# p__Crenarchaeota__c__Thermoprotei
for prefix in Taxonomy.rank_prefixes:
split_str = '__' + prefix
if split_str in taxa_str:
taxa_str = taxa_str.replace(
split_str, ';' + prefix)
# appears to be an internal label and not simply a
# support value
taxa = [x.strip() for x in taxa_str.split(';')] + taxa
node = node.parent_node
if warnings and len(taxa) > 7:
self.logger.warning(
'Invalid taxonomy string read from tree for taxon %s: %s' %
(leaf.taxon.label, taxa))
# sys.exit(-1)
# check if genus name should be appended to species label
if len(taxa) == 7:
genus = taxa[5][3:]
species = taxa[6][3:]
if genus not in species:
taxa[6] = 's__' + genus + ' ' + species
taxa = self.fill_trailing_ranks(taxa)
taxonomy[leaf.taxon.label] = taxa
return taxonomy