def write_smc(filename, smc):
"""Writes a SMC file"""
out = argweaver.open_stream(filename, "w")
for item in smc:
if item["tag"] == "NAMES":
util.print_row("NAMES", *item["names"], out=out)
elif item["tag"] == "REGION":
util.print_row("REGION",
item["chrom"],
item["start"],
item["end"],
out=out)
elif item["tag"] == "TREE":
if not isinstance(item["tree"], basestring):
tree = format_tree(item["tree"])
else:
tree = item["tree"]
util.print_row("TREE", item["start"], item["end"], tree, out=out)
elif item["tag"] == "SPR":
util.print_row("SPR",
item["pos"],
item["recomb_node"],
item["recomb_time"],
item["coal_node"],
item["coal_time"],
out=out)
out.close()
def write_smc(filename, smc):
"""Writes a SMC file"""
out = argweaver.open_stream(filename, "w")
for item in smc:
if item["tag"] == "NAMES":
util.print_row("NAMES", *item["names"], out=out)
elif item["tag"] == "REGION":
util.print_row("REGION",
item["chrom"], item["start"], item["end"], out=out)
elif item["tag"] == "TREE":
if not isinstance(item["tree"], basestring):
tree = format_tree(item["tree"])
else:
tree = item["tree"]
util.print_row("TREE", item["start"], item["end"], tree, out=out)
elif item["tag"] == "SPR":
util.print_row("SPR", item["pos"],
item["recomb_node"], item["recomb_time"],
item["coal_node"], item["coal_time"], out=out)
out.close()
def iter_arg_layout(filename):
"""
Iterate through an ARG layout file.
"""
with closing(argweaver.open_stream(filename, compress='bgzip')) as infile:
for line in infile:
tokens = line.rstrip().split("\t")
block = [tokens[0], int(tokens[1]), int(tokens[2])]
leaf_layout = {}
for i in range(3, len(tokens), 2):
leaf_layout[tokens[i]] = float(tokens[i + 1])
yield block, leaf_layout
def iter_arg_layout(filename):
"""
Iterate through an ARG layout file.
"""
with closing(argweaver.open_stream(filename, compress='bgzip')) as infile:
for line in infile:
tokens = line.rstrip().split("\t")
block = [tokens[0], int(tokens[1]), int(tokens[2])]
leaf_layout = {}
for i in range(3, len(tokens), 2):
leaf_layout[tokens[i]] = float(tokens[i+1])
yield block, leaf_layout
def iter_smc_file(filename, parse_trees=False, apply_spr=False, region=None):
"""
Iterates through a SMC file.
parse_trees: If True, parses local trees.
apply_spr: If True, avoids reading each tree by applying the SPR
operation to the current tree.
region: If given, returns only trees and SPRs within region=(start, end).
Yields item, where item can be one of the following:
{'tag': 'NAMES',
'names': names_of_sequences}
{'tag': 'REGION',
'chrom': name_of_chromosome,
'start': start_coordinate_of_region,
'end': end_coordinate_of_region}
{'tag': 'TREE',
'start': start_coordinate_of_local_region,
'end': end_coordinate_of_local_region,
'tree': local_tree}
{'tag': 'SPR',
'pos': coordinate of recombination point,
'recomb_node': name_of_recombination_node,
'recomb_time': time_of_recombination,
'coal_node': name_of_branch_with_recoalescence,
'coal_time': time_of_recoalescence}
"""
if region:
tree = None
spr = None
for item in iter_subsmc(iter_smc_file(filename), region):
if item["tag"] == "SPR":
spr = item
elif item["tag"] == "TREE":
if parse_trees:
if apply_spr and tree is not None and spr is not None:
smc_apply_spr(tree, spr)
else:
tree = parse_tree(item["tree"])
item["tree"] = tree
yield item
return
with closing(argweaver.open_stream(filename)) as infile:
spr = None
tree = None
for line in infile:
line = line.rstrip()
tokens = line.split("\t")
if tokens[0] == "NAMES":
yield {"tag": "NAMES", "names": tokens[1:]}
elif tokens[0] == "REGION":
yield {
"tag": "REGION",
"chrom": tokens[1],
"start": int(tokens[2]),
"end": int(tokens[3])
}
elif tokens[0] == "RANGE":
raise Exception("deprecated RANGE line, use REGION instead")
elif tokens[0] == "TREE":
tree_text = tokens[3]
if parse_trees:
if apply_spr and tree is not None and spr is not None:
smc_apply_spr(tree, spr)
else:
tree = parse_tree(tree_text)
else:
tree = tree_text
yield {
"tag": "TREE",
"start": int(tokens[1]),
"end": int(tokens[2]),
"tree": tree
}
elif tokens[0] == "SPR":
spr = {
"tag": "SPR",
"pos": int(tokens[1]),
"recomb_node": int(tokens[2]),
"recomb_time": float(tokens[3]),
"coal_node": int(tokens[4]),
"coal_time": float(tokens[5])
}
yield spr
def iter_smc_file(filename, parse_trees=False, apply_spr=False,
region=None):
"""
Iterates through a SMC file.
parse_trees: If True, parses local trees.
apply_spr: If True, avoids reading each tree by applying the SPR
operation to the current tree.
region: If given, returns only trees and SPRs within region=(start, end).
Yields item, where item can be one of the following:
{'tag': 'NAMES',
'names': names_of_sequences}
{'tag': 'REGION',
'chrom': name_of_chromosome,
'start': start_coordinate_of_region,
'end': end_coordinate_of_region}
{'tag': 'TREE',
'start': start_coordinate_of_local_region,
'end': end_coordinate_of_local_region,
'tree': local_tree}
{'tag': 'SPR',
'pos': coordinate of recombination point,
'recomb_node': name_of_recombination_node,
'recomb_time': time_of_recombination,
'coal_node': name_of_branch_with_recoalescence,
'coal_time': time_of_recoalescence}
"""
if region:
tree = None
spr = None
for item in iter_subsmc(iter_smc_file(filename), region):
if item["tag"] == "SPR":
spr = item
elif item["tag"] == "TREE":
if parse_trees:
if apply_spr and tree is not None and spr is not None:
smc_apply_spr(tree, spr)
else:
tree = parse_tree(item["tree"])
item["tree"] = tree
yield item
return
with closing(argweaver.open_stream(filename)) as infile:
spr = None
tree = None
for line in infile:
line = line.rstrip()
tokens = line.split("\t")
if tokens[0] == "NAMES":
yield {"tag": "NAMES", "names": tokens[1:]}
elif tokens[0] == "REGION":
yield {"tag": "REGION",
"chrom": tokens[1],
"start": int(tokens[2]),
"end": int(tokens[3])}
elif tokens[0] == "RANGE":
raise Exception("deprecated RANGE line, use REGION instead")
elif tokens[0] == "TREE":
tree_text = tokens[3]
if parse_trees:
if apply_spr and tree is not None and spr is not None:
smc_apply_spr(tree, spr)
else:
tree = parse_tree(tree_text)
else:
tree = tree_text
yield {"tag": "TREE",
"start": int(tokens[1]),
"end": int(tokens[2]),
"tree": tree}
elif tokens[0] == "SPR":
spr = {"tag": "SPR",
"pos": int(tokens[1]),
"recomb_node": int(tokens[2]),
"recomb_time": float(tokens[3]),
"coal_node": int(tokens[4]),
"coal_time": float(tokens[5])}
yield spr