rename-nodes.py

#!/usr/bin/env python

from __future__ import print_function

from ete3 import Tree
import sys
import re

numeric_prefix = re.compile('^\d+_')
numeric = re.compile('^\d+$')

t = Tree(sys.argv[1], format=1)

for node in t.traverse():
    match = numeric_prefix.match(node.name)
    if match:
        # print('%s has a numeric prefix' % node.name, file=sys.stderr)
        # print('%s -> %s' % (node.name, node.name[match.end():]),
        # file=sys.stderr)
        node.name = node.name[match.end():]
    else:
        match = numeric.match(node.name)
        if match:
            # print('%s is all digits' % node.name, file=sys.stderr)
            node.name = 'ancestral-node-' + node.name
        else:
            raise ValueError(
                "Tree node label '%s' has an unrecognized format." % node.name)

# Print the tree, including the name of the root.
print('%s%s;' % (t.write(format=1)[:-1], t.name))