def get_parent_tip_ranks(self, tax_tree):
rank_tips = {}
rank_parent = {}
for node in tax_tree.traverse("postorder"):
if node.is_leaf() or node.is_root():
continue
tax_path = node.name
ranks = Taxonomy.split_rank_uid(tax_path)
rank_lvl = Taxonomy.lowest_assigned_rank_level(ranks)
if rank_lvl < 2:
continue
parent_ranks = Taxonomy.split_rank_uid(node.up.name)
parent_lvl = Taxonomy.lowest_assigned_rank_level(parent_ranks)
if parent_lvl < 1:
continue
rank_seqs = node.get_leaf_names()
rank_size = len(rank_seqs)
if rank_size < 2 or rank_size > self.reftree_size-4:
continue
# print rank_lvl, "\t", tax_path, "\t", rank_seqs, "\n"
rank_tips[tax_path] = node.get_leaf_names()
rank_parent[tax_path] = parent_ranks
return rank_parent, rank_tips
def get_parent_tip_ranks(self, tax_tree):
rank_tips = {}
rank_parent = {}
for node in tax_tree.traverse("postorder"):
if node.is_leaf() or node.is_root():
continue
tax_path = node.name
ranks = Taxonomy.split_rank_uid(tax_path)
rank_lvl = Taxonomy.lowest_assigned_rank_level(ranks)
if rank_lvl < 2:
continue
parent_ranks = Taxonomy.split_rank_uid(node.up.name)
parent_lvl = Taxonomy.lowest_assigned_rank_level(parent_ranks)
if parent_lvl < 1:
continue
rank_seqs = node.get_leaf_names()
rank_size = len(rank_seqs)
if rank_size < 2 or rank_size > self.reftree_size - 4:
continue
# print rank_lvl, "\t", tax_path, "\t", rank_seqs, "\n"
rank_tips[tax_path] = node.get_leaf_names()
rank_parent[tax_path] = parent_ranks
return rank_parent, rank_tips
def check_seq_tax_labels(self, seq_name, orig_ranks, ranks, lws):
mis_rec = None
num_common_ranks = len(self.tax_common_ranks)
orig_rank_level = Taxonomy.lowest_assigned_rank_level(orig_ranks)
new_rank_level = Taxonomy.lowest_assigned_rank_level(ranks)
#if new_rank_level < 0 or (new_rank_level < num_common_ranks and orig_rank_level >= num_common_ranks):
# if new_rank_level < 0:
if len(ranks) == 0:
mis_rec = {}
mis_rec['name'] = seq_name
mis_rec['orig_level'] = -1
mis_rec['real_level'] = 0
mis_rec['level_name'] = "[NotIngroup]"
mis_rec['inv_level'] = -1 * mis_rec[
'real_level'] # just for sorting
mis_rec['orig_ranks'] = orig_ranks
mis_rec['ranks'] = []
mis_rec['lws'] = [1.0]
mis_rec['conf'] = mis_rec['lws'][0]
else:
mislabel_lvl = -1
min_len = min(len(orig_ranks), len(ranks))
for rank_lvl in range(min_len):
if ranks[rank_lvl] != Taxonomy.EMPTY_RANK and ranks[
rank_lvl] != orig_ranks[rank_lvl]:
mislabel_lvl = rank_lvl
break
if mislabel_lvl >= 0:
real_lvl = self.tax_code.guess_rank_level(
orig_ranks, mislabel_lvl)
mis_rec = {}
mis_rec['name'] = seq_name
mis_rec['orig_level'] = mislabel_lvl
mis_rec['real_level'] = real_lvl
mis_rec['level_name'] = self.tax_code.rank_level_name(
real_lvl)[0]
mis_rec['inv_level'] = -1 * mis_rec[
'real_level'] # just for sorting
mis_rec['orig_ranks'] = orig_ranks
mis_rec['ranks'] = ranks
mis_rec['lws'] = lws
mis_rec['conf'] = lws[mislabel_lvl]
if mis_rec:
self.mislabels.append(mis_rec)
return mis_rec
def label_bf_tree_with_ranks(self):
"""labeling inner tree nodes with taxonomic ranks"""
if not self.bf_rooted_tree:
raise AssertionError(
"self.bf_rooted_tree is not set: TaxTreeHelper.set_bf_unrooted_tree() must be called before!"
)
for node in self.bf_rooted_tree.traverse("postorder"):
if node.is_leaf():
seq_ranks = self.origin_taxonomy[node.name]
rank_level = Taxonomy.lowest_assigned_rank_level(seq_ranks)
node.add_feature("rank_level", rank_level)
node.add_feature("ranks", seq_ranks)
node.name += "__" + seq_ranks[rank_level]
else:
if len(node.children) != 2:
raise AssertionError("FATAL ERROR: tree is not bifurcating!")
lchild = node.children[0]
rchild = node.children[1]
rank_level = min(lchild.rank_level, rchild.rank_level)
while rank_level >= 0 and lchild.ranks[rank_level] != rchild.ranks[rank_level]:
rank_level -= 1
node.add_feature("rank_level", rank_level)
node_ranks = [Taxonomy.EMPTY_RANK] * 7
if rank_level >= 0:
node_ranks[0 : rank_level + 1] = lchild.ranks[0 : rank_level + 1]
node.name = lchild.ranks[rank_level]
else:
node.name = "Undefined"
if hasattr(node, "B") and self.cfg.verbose:
print "INFO: no taxonomic annotation for branch %s (reason: children belong to different kingdoms)" % node.B
node.add_feature("ranks", node_ranks)
self.tax_tree = self.bf_rooted_tree
def check_seq_tax_labels(self, seq_name, orig_ranks, ranks, lws):
mis_rec = None
num_common_ranks = len(self.tax_common_ranks)
orig_rank_level = Taxonomy.lowest_assigned_rank_level(orig_ranks)
new_rank_level = Taxonomy.lowest_assigned_rank_level(ranks)
#if new_rank_level < 0 or (new_rank_level < num_common_ranks and orig_rank_level >= num_common_ranks):
# if new_rank_level < 0:
if len(ranks) == 0:
mis_rec = {}
mis_rec['name'] = seq_name
mis_rec['orig_level'] = -1
mis_rec['real_level'] = 0
mis_rec['level_name'] = "[NotIngroup]"
mis_rec['inv_level'] = -1 * mis_rec['real_level'] # just for sorting
mis_rec['orig_ranks'] = orig_ranks
mis_rec['ranks'] = []
mis_rec['lws'] = [1.0]
mis_rec['conf'] = mis_rec['lws'][0]
else:
mislabel_lvl = -1
min_len = min(len(orig_ranks),len(ranks))
for rank_lvl in range(min_len):
if ranks[rank_lvl] != Taxonomy.EMPTY_RANK and ranks[rank_lvl] != orig_ranks[rank_lvl]:
mislabel_lvl = rank_lvl
break
if mislabel_lvl >= 0:
real_lvl = self.tax_code.guess_rank_level(orig_ranks, mislabel_lvl)
mis_rec = {}
mis_rec['name'] = seq_name
mis_rec['orig_level'] = mislabel_lvl
mis_rec['real_level'] = real_lvl
mis_rec['level_name'] = self.tax_code.rank_level_name(real_lvl)[0]
mis_rec['inv_level'] = -1 * mis_rec['real_level'] # just for sorting
mis_rec['orig_ranks'] = orig_ranks
mis_rec['ranks'] = ranks
mis_rec['lws'] = lws
mis_rec['conf'] = lws[mislabel_lvl]
if mis_rec:
self.mislabels.append(mis_rec)
return mis_rec
def run_leave_subtree_out_test(self):
job_name = self.cfg.subst_name("l1out_rank_%NAME%")
# if self.jplace_fname:
# jp = EpaJsonParser(self.jplace_fname)
# else:
#create file with subtrees
rank_parent, rank_tips = self.get_parent_tip_ranks(self.tax_tree)
subtree_list = list(rank_tips.items())
if len(subtree_list) == 0:
return 0
subtree_list_file = self.cfg.tmp_fname("treelist_%NAME%.txt")
with open(subtree_list_file, "w") as fout:
for rank_name, tips in subtree_list:
fout.write("%s\n" % " ".join(tips))
jp_list = self.raxml.run_epa(job_name,
self.refalign_fname,
self.reftree_fname,
self.optmod_fname,
mode="l1o_subtree",
subtree_fname=subtree_list_file)
subtree_count = 0
for jp in jp_list:
placements = jp.get_placement()
for place in placements:
ranks, lws = self.classify_seq(place)
tax_path = subtree_list[subtree_count][0]
orig_ranks = Taxonomy.split_rank_uid(tax_path)
rank_level = Taxonomy.lowest_assigned_rank_level(orig_ranks)
rank_prefix = self.tax_code.guess_rank_level_name(
orig_ranks, rank_level)[0]
rank_name = orig_ranks[rank_level]
if not rank_name.startswith(rank_prefix):
rank_name = rank_prefix + rank_name
parent_ranks = rank_parent[tax_path]
# print orig_ranks, "\n", parent_ranks, "\n", ranks, "\n"
mis_rec = self.check_rank_tax_labels(rank_name, parent_ranks,
ranks, lws)
if mis_rec:
self.misrank_conf_map[tax_path] = mis_rec['conf']
subtree_count += 1
return subtree_count
def run_leave_subtree_out_test(self):
job_name = self.cfg.subst_name("l1out_rank_%NAME%")
# if self.jplace_fname:
# jp = EpaJsonParser(self.jplace_fname)
# else:
#create file with subtrees
rank_parent, rank_tips = get_parent_tip_ranks(self.tax_tree)
subtree_list = rank_tips.items()
if len(subtree_list) == 0:
return 0
subtree_list_file = self.cfg.tmp_fname("treelist_%NAME%.txt")
with open(subtree_list_file, "w") as fout:
for rank_name, tips in subtree_list:
fout.write("%s\n" % " ".join(tips))
jp_list = self.raxml.run_epa(job_name, self.refalign_fname, self.reftree_fname, self.optmod_fname,
mode="l1o_subtree", subtree_fname=subtree_list_file)
subtree_count = 0
for jp in jp_list:
placements = jp.get_placement()
for place in placements:
ranks, lws = self.classify_seq(place)
tax_path = subtree_list[subtree_count][0]
orig_ranks = Taxonomy.split_rank_uid(tax_path)
rank_level = Taxonomy.lowest_assigned_rank_level(orig_ranks)
rank_prefix = self.guess_rank_level_name(orig_ranks, rank_level)[0]
rank_name = orig_ranks[rank_level]
if not rank_name.startswith(rank_prefix):
rank_name = rank_prefix + rank_name
parent_ranks = rank_parent[tax_path]
# print orig_ranks, "\n", parent_ranks, "\n", ranks, "\n"
mis_rec = self.check_rank_tax_labels(rank_name, parent_ranks, ranks, lws)
if mis_rec:
self.misrank_conf_map[tax_path] = mis_rec['conf']
subtree_count += 1
return subtree_count
def run_leave_subtree_out_test(self):
job_name = self.cfg.subst_name("l1out_rank_%NAME%")
# if self.jplace_fname:
# jp = EpaJsonParser(self.jplace_fname)
# else:
#create file with subtrees
rank_tips = {}
rank_parent = {}
for node in self.tax_tree.traverse("postorder"):
if node.is_leaf() or node.is_root():
continue
tax_path = node.name
ranks = Taxonomy.split_rank_uid(tax_path)
rank_lvl = Taxonomy.lowest_assigned_rank_level(ranks)
if rank_lvl < 2:
continue
parent_ranks = Taxonomy.split_rank_uid(node.up.name)
parent_lvl = Taxonomy.lowest_assigned_rank_level(parent_ranks)
if parent_lvl < 1:
continue
rank_seqs = node.get_leaf_names()
rank_size = len(rank_seqs)
if rank_size < 2 or rank_size > self.reftree_size-4:
continue
# print rank_lvl, "\t", tax_path, "\t", rank_seqs, "\n"
rank_tips[tax_path] = node.get_leaf_names()
rank_parent[tax_path] = parent_ranks
subtree_list = rank_tips.items()
if len(subtree_list) == 0:
return 0
subtree_list_file = self.cfg.tmp_fname("treelist_%NAME%.txt")
with open(subtree_list_file, "w") as fout:
for rank_name, tips in subtree_list:
fout.write("%s\n" % " ".join(tips))
jp_list = self.raxml.run_epa(job_name, self.refalign_fname, self.reftree_fname, self.optmod_fname,
mode="l1o_subtree", subtree_fname=subtree_list_file)
subtree_count = 0
for jp in jp_list:
placements = jp.get_placement()
for place in placements:
ranks, lws = self.classify_seq(place)
tax_path = subtree_list[subtree_count][0]
orig_ranks = Taxonomy.split_rank_uid(tax_path)
rank_level = Taxonomy.lowest_assigned_rank_level(orig_ranks)
rank_prefix = self.guess_rank_level_name(orig_ranks, rank_level)[0]
rank_name = orig_ranks[rank_level]
if not rank_name.startswith(rank_prefix):
rank_name = rank_prefix + rank_name
parent_ranks = rank_parent[tax_path]
# print orig_ranks, "\n", parent_ranks, "\n", ranks, "\n"
mis_rec = self.check_rank_tax_labels(rank_name, parent_ranks, ranks, lws)
if mis_rec:
self.misrank_conf_map[tax_path] = mis_rec['conf']
subtree_count += 1
return subtree_count
def run_leave_subtree_out_test(self):
job_name = self.cfg.subst_name("l1out_rank_%NAME%")
# if self.jplace_fname:
# jp = EpaJsonParser(self.jplace_fname)
# else:
#create file with subtrees
rank_tips = {}
rank_parent = {}
for node in self.tax_tree.traverse("postorder"):
if node.is_leaf() or node.is_root():
continue
tax_path = node.name
ranks = Taxonomy.split_rank_uid(tax_path)
rank_lvl = Taxonomy.lowest_assigned_rank_level(ranks)
if rank_lvl < 2:
continue
parent_ranks = Taxonomy.split_rank_uid(node.up.name)
parent_lvl = Taxonomy.lowest_assigned_rank_level(parent_ranks)
if parent_lvl < 1:
continue
rank_seqs = node.get_leaf_names()
rank_size = len(rank_seqs)
if rank_size < 2 or rank_size > self.reftree_size - 4:
continue
# print rank_lvl, "\t", tax_path, "\t", rank_seqs, "\n"
rank_tips[tax_path] = node.get_leaf_names()
rank_parent[tax_path] = parent_ranks
subtree_list = rank_tips.items()
if len(subtree_list) == 0:
return 0
subtree_list_file = self.cfg.tmp_fname("treelist_%NAME%.txt")
with open(subtree_list_file, "w") as fout:
for rank_name, tips in subtree_list:
fout.write("%s\n" % " ".join(tips))
jp_list = self.raxml.run_epa(job_name,
self.refalign_fname,
self.reftree_fname,
self.optmod_fname,
mode="l1o_subtree",
subtree_fname=subtree_list_file)
subtree_count = 0
for jp in jp_list:
placements = jp.get_placement()
for place in placements:
ranks, lws = self.classify_seq(place)
tax_path = subtree_list[subtree_count][0]
orig_ranks = Taxonomy.split_rank_uid(tax_path)
rank_level = Taxonomy.lowest_assigned_rank_level(orig_ranks)
rank_prefix = self.guess_rank_level_name(
orig_ranks, rank_level)[0]
rank_name = orig_ranks[rank_level]
if not rank_name.startswith(rank_prefix):
rank_name = rank_prefix + rank_name
parent_ranks = rank_parent[tax_path]
# print orig_ranks, "\n", parent_ranks, "\n", ranks, "\n"
mis_rec = self.check_rank_tax_labels(rank_name, parent_ranks,
ranks, lws)
if mis_rec:
self.misrank_conf_map[tax_path] = mis_rec['conf']
subtree_count += 1
return subtree_count
