suppresses nodes from an OT tree if they do not have an ID listed in a separate file. Takes newick-tree-filepath and file-with-ids-one-per-line as args

suppress-non-listed-ids-or-unnamed.py

#!/bin/env python3
import sys
import dendropy
import re
id_extractor = re.compile('.*ott([0-9]+)$')
tree_filepath = sys.argv[1]
id_filepath = sys.argv[2]
tree = dendropy.Tree.get(path=tree_filepath, schema='newick', suppress_internal_node_taxa=False)

# Get the list of IDs to retain
trl = []
for i in open(id_filepath, 'r').readlines():
    if i.lower().startswith('ott'):
        i = i[3:]
    i = i.strip()
    while i.endswith(','):
        i = i[:-1]
    trl.append(i)
to_retain_ids = frozenset(trl)

to_suppress = []
for node in tree.postorder_node_iter():
    if node.taxon:
        m = id_extractor.match(node.taxon.label)
        if not m:
            sys.exit('name not matching pattern: {}'.format(node.taxon.label))
        id_for_node = m.group(1)
        if id_for_node not in to_retain_ids:
            sys.stderr.write('flagging for suppression due to non-retained ID: {}\n'.format(node.taxon.label))
            to_suppress.append(node)
    else:
        assert(node.num_child_nodes() != 0)
        to_suppress.append(node)

for node in to_suppress:
    if node.edge:
        node.edge.collapse()

tree.write_to_stream(sys.stdout, schema="newick")