def decompose_trees(tree, max_subset_size):
"""
"""
next_trees = [tree]
done_trees = []
while len(next_trees) > 0:
trees = next_trees
next_trees = []
for tree in trees:
t1, t2 = bisect_tree(tree,
breaking_edge_style="centroid")
n1 = t1.n_leaves
n2 = t2.n_leaves
if n1 > max_subset_size:
next_trees.append(t1)
else:
if n1 >= 5:
done_trees.append(t1)
else:
sys.exit("T1 has fewer than 5 leaves!")
if n2 > max_subset_size:
next_trees.append(t2)
else:
if n2 >= 5:
done_trees.append(t2)
else:
sys.exit("T2 has fewer than 5 leaves!")
return done_trees
def _do_test_longest(self, t, level="1"):
if t.n_leaves < 3:
return
before_br_len = [
e.length for e in t._tree.preorder_edge_iter() if e.length
]
_LOG.debug("code=%s\n before = %s" % (level, t.compose_newick()))
_LOG.debug(" after len(before_br_len) = %d" % (len(before_br_len)))
num_real_edges_before = len(before_br_len)
if len(t._tree.seed_node.child_nodes()) < 3:
num_real_edges_before -= 1
t1, t2 = bisect_tree(t, 'longest')
after_1_br_len = [
e.length for e in t1._tree.preorder_edge_iter() if e.length
]
after_2_br_len = [
e.length for e in t2._tree.preorder_edge_iter() if e.length
]
num_branches_1 = len(after_1_br_len)
num_branches_2 = len(after_2_br_len)
if num_branches_2 == 0:
num_branches_2 = 1
if num_branches_1 == 0:
num_branches_1 = 1
expected_diff = 3
if num_branches_2 == 1:
expected_diff -= 2
if num_branches_1 == 1:
expected_diff -= 2
# cherries are rooted, so they make 1 edge look like 2
if len(t1._tree.seed_node.child_nodes()) == 2:
num_branches_1 -= 1
if len(t2._tree.seed_node.child_nodes()) == 2:
num_branches_2 -= 1
_LOG.debug(" after 1 = %s" % (t1.compose_newick()))
_LOG.debug(" after num_branches_1 = %d" % (num_branches_1))
_LOG.debug(" after 2 = %s" % (t2.compose_newick()))
_LOG.debug(" after num_branches_2 = %d" % (num_branches_2))
#self.assertEqual(len(before_br_len), expected_diff + num_branches_1 + num_branches_2)
before_br_len.sort(reverse=True)
before_br_len.pop(0)
before_sum = sum(before_br_len)
after_sum = sum(after_1_br_len) + sum(after_2_br_len)
diff = before_sum - after_sum
self.assertTrue(abs(diff) < TOL)
if t1.n_leaves > 2:
nl = level + ".1"
self._do_test_longest(t1, level=nl)
if t2.n_leaves > 2:
nl = level + ".2"
self._do_test_longest(t2, level=nl)
def _do_test_longest(self, t, level="1"):
if t.n_leaves < 3:
return
before_br_len = [e.length for e in t._tree.preorder_edge_iter() if e.length]
_LOG.debug("code=%s\n before = %s" % (level, t.compose_newick()))
_LOG.debug(" after len(before_br_len) = %d" % (len(before_br_len)))
num_real_edges_before = len(before_br_len)
if len(t._tree.seed_node.child_nodes()) < 3:
num_real_edges_before -= 1
t1, t2 = bisect_tree(t, 'longest')
after_1_br_len = [e.length for e in t1._tree.preorder_edge_iter() if e.length]
after_2_br_len = [e.length for e in t2._tree.preorder_edge_iter() if e.length]
num_branches_1 = len(after_1_br_len)
num_branches_2 = len(after_2_br_len)
if num_branches_2 == 0:
num_branches_2 = 1
if num_branches_1 == 0:
num_branches_1 = 1
expected_diff = 3
if num_branches_2 == 1:
expected_diff -= 2
if num_branches_1 == 1:
expected_diff -= 2
# cherries are rooted, so they make 1 edge look like 2
if len(t1._tree.seed_node.child_nodes()) == 2:
num_branches_1 -= 1
if len(t2._tree.seed_node.child_nodes()) == 2:
num_branches_2 -= 1
_LOG.debug(" after 1 = %s" % (t1.compose_newick()))
_LOG.debug(" after num_branches_1 = %d" % (num_branches_1))
_LOG.debug(" after 2 = %s" % (t2.compose_newick()))
_LOG.debug(" after num_branches_2 = %d" % (num_branches_2))
#self.assertEqual(len(before_br_len), expected_diff + num_branches_1 + num_branches_2)
before_br_len.sort(reverse=True)
before_br_len.pop(0)
before_sum = sum(before_br_len)
after_sum = sum(after_1_br_len) + sum(after_2_br_len)
diff = before_sum - after_sum
self.assertTrue(abs(diff) < TOL)
if t1.n_leaves > 2:
nl = level+ ".1"
self._do_test_longest(t1, level=nl)
if t2.n_leaves > 2:
nl = level+ ".2"
self._do_test_longest(t2, level=nl)
def decompose_trees(tree, max_subset_size):
"""
"""
next_trees = [tree]
done_trees = []
while len(next_trees) > 0:
trees = next_trees
next_trees = []
for tree in trees:
t1, t2 = bisect_tree(tree, breaking_edge_style="centroid")
no1 = t1.n_leaves
no2 = t2.n_leaves
d_t1 = t1._tree
d_t2 = t2._tree
d_t1_mat = d_t1.phylogenetic_distance_matrix()
d_t2_mat = d_t2.phylogenetic_distance_matrix()
diam_t1 = -1
diam_t2 = -1
for i, n1 in enumerate(d_t1.taxon_namespace[:-1]):
for n2 in d_t1.taxon_namespace[i + 1:]:
diam_t1 = max(diam_t1, d_t1_mat(n1, n2))
for i, n1 in enumerate(d_t2.taxon_namespace[:-1]):
for n2 in d_t2.taxon_namespace[i + 1:]:
diam_t2 = max(diam_t2, d_t2_mat(n1, n2))
if no1 > max_subset_size:
next_trees.append(t1)
else:
if no1 >= 5:
done_trees.append(t1)
# print(diam_t1)
else:
sys.exit("T1 has fewer than 5 leaves!")
if no2 > max_subset_size:
next_trees.append(t2)
else:
if no2 >= 5:
done_trees.append(t2)
# print(diam_t2)
else:
sys.exit("T2 has fewer than 5 leaves!")
return done_trees
def decompose_phylogeny(phy, max_size, min_size):
trees_map = []
tree_list = [phy]
while len(tree_list) > 0:
tmp_phy = tree_list.pop()
t1, t2 = bisect_tree(tree=tmp_phy,
breaking_edge_style='midpoint',
max_size=max_size)
if t1.count_leaves() > min_size:
tree_list.append(deepcopy(t1))
else:
trees_map.append(deepcopy(t1))
if t2.count_leaves() > min_size:
tree_list.append(deepcopy(t2))
else:
trees_map.append(deepcopy(t2))
return trees_map
def main(args):
# Step 1: Decompose tree
tree = dendropy.Tree.get(path=args.input_tree_file, schema="newick")
tree.resolve_polytomies(limit=2, update_bipartitions=True)
tree = PhylogeneticTree(tree)
t1, t2 = bisect_tree(tree)
trees = [t1, t2]
# Step 2: Write out leaf subsets
# i = 1
i = 0
keep1 = t1.leaf_node_names()
with open(args.output + "/A.lab", "w") as f:
f.write("\n".join(keep1))
keep2 = t2.leaf_node_names()
with open(args.output + "/B.lab", "w") as f:
f.write("\n".join(keep2))
def _do_test_centroid(self, t, level="1"):
if t.n_leaves < 5:
return
t.calc_splits()
t1, t2 = bisect_tree(t, 'centroid')
assert t1.n_leaves + t2.n_leaves == t.n_leaves
# indent = level.count(".")
# print("==============\nInput tree has %s leaf nodes." % t.n_leaves)
# print("Subtree 1 tree has %s leaf nodes." % t1.n_leaves)
# print("Subtree 2 tree has %s leaf nodes." % t2.n_leaves)
# print("==============\n")
if t1.n_leaves > 2:
nl = level + ".1"
self._do_test_centroid(t1, level=nl)
if t2.n_leaves > 2:
nl = level + ".2"
self._do_test_centroid(t2, level=nl)
def _do_test_centroid(self, t, level="1"):
if t.n_leaves < 5:
return
t.calc_splits()
t1, t2 = bisect_tree(t, 'centroid')
assert t1.n_leaves + t2.n_leaves == t.n_leaves
# indent = level.count(".")
# print("==============\nInput tree has %s leaf nodes." % t.n_leaves)
# print("Subtree 1 tree has %s leaf nodes." % t1.n_leaves)
# print("Subtree 2 tree has %s leaf nodes." % t2.n_leaves)
# print("==============\n")
if t1.n_leaves > 2:
nl = level+ ".1"
self._do_test_centroid(t1, level=nl)
if t2.n_leaves > 2:
nl = level+ ".2"
self._do_test_centroid(t2, level=nl)