def load_parameters():
descendant_dict = defaultdict()
with open(input_file, 'r') as f:
all_lines = f.readlines()
aln_file = all_lines[0].strip()
if not os.path.exists(aln_file):
logger.error('Invalid cDNA alignment file: {0}'.format(aln_file))
sys.exit(1)
logger.info('Input cDNA alignment file: {0}'.format(aln_file))
seq_id_dict = defaultdict()
seq_id_list = []
for seq_record in AlignIO.read(aln_file, 'fasta'):
seq_id_dict[str(seq_record.id)] = 1
seq_id_list.append(str(seq_record.id))
tree_file = all_lines[1].strip()
if not os.path.exists(tree_file):
logger.error('Invalid tree file: {0}'.format(tree_file))
sys.exit(1)
logger.info('Input tree file: {0}'.format(tree_file))
tmp_t = Tree(tree_file, format=0)
node_id_dict = defaultdict()
for node in tmp_t:
node_id_dict[str(node.name)] = 1
if seq_id_dict != node_id_dict:
if len(seq_id_dict) < len(node_id_dict):
logger.warning('Sequences is less than tree nodes.')
logger.info('Trim input tree file.')
tree_file = trim_tree(tree_file, seq_id_list)
else:
logger.error('Sequences is falsely greater than tree nodes.')
sys.exit(1)
t = EvolTree(tree_file, format=1)
for descendant in t.iter_descendants():
descendant_dict[descendant.node_id] = str(descendant)
root = t.get_tree_root()
id_list = []
for leaf in t.traverse('preorder'):
id_list.append(leaf.node_id)
select_nodes = []
if len(all_lines) > 2:
for each_line in all_lines[2:]:
s = each_line.strip()
if s:
select_nodes.append(s)
if select_nodes:
nodes_line = ', '.join(select_nodes)
logger.info('Input nodes: {0}'.format(nodes_line))
for node in select_nodes:
if node not in t:
logger.error('Error node: {0}'.format(node))
sys.exit(1)
if not t.check_monophyly(values=select_nodes, target_attr='name'):
logger.error('Some nodes are not monophyletic.')
sys.exit(1)
common_ancestor = t.get_common_ancestor(select_nodes)
else:
common_ancestor = root
logger.info('No specific node')
run_list = []
for s in common_ancestor.iter_descendants():
run_list.append(s.node_id)
logger.info('These node ids will be checked: {0}'.format(
str(run_list)))
return run_list, aln_file, tree_file, descendant_dict
