def _label(tree, ancestors):
"""
Relabel internal nodes of a tree and map them to the corresponding name of
ancestral sequences.
:param tree: str, a NEWICK format string or file for a tree (must start with
"(" and end with ';').
:param ancestors: dict, a dict object stores sequences.
:return: tuple, a relabeled tree object and a dict object for sequences.
"""
if isinstance(tree, str):
if os.path.isfile(tree):
pass
elif tree.startswith('(') and tree.endswith(';'):
tree = StringIO(tree)
else:
error('Invalid tree encounter, tree relabel aborted.')
sys.exit(1)
tree = Phylo.read(tree, 'newick')
number, maps = tree.count_terminals(), {}
for clade in tree.find_clades():
if not clade.is_terminal():
clade.confidence = None
number += 1
old, new = clade.name, 'NODE{}'.format(number)
maps[old] = new
clade.name = new
ancestors = {maps.get(k, k): v for k, v in ancestors.items()}
return tree, ancestors
def __init__(self, tree, leave=False):
"""
A class handles phylogenetic trees.
:param tree: str, a newick tree file or tree string (must start with
'(' and end with ';').
:param leave: bool, whether exit the process or not once an error
occurred.
"""
if isinstance(tree, str):
if tree.startswith('(') and tree.endswith(';'):
tree = StringIO(tree)
elif os.path.isfile(tree):
pass
else:
error(
'Invalid tree: {}, tree should be either a NEWICK format '
'tree string or tree file.'.format(tree))
tree = None
else:
error('Invalid tree, tree should be a string.')
tree = None
if tree is None and leave:
sys.exit(1)
else:
tree = Phylo.read(tree, 'newick') if tree else None
if tree:
leaves = len([c for c in tree.find_clades() if c.is_terminal()])
nodes = len([c for c in tree.find_clades() if not c.is_terminal()])
length = sum([
c.branch_length if c.branch_length else 0.0
for c in tree.find_clades()
])
else:
leaves, nodes, length = 0, 0, 0.0
self.tree = tree
self.leaves = leaves
self.nodes = nodes
self.length = length
