def prophyle_decompress(archive, output_dir, klcp):
pro.test_files(archive)
_compile_prophyle_bin(parallel=True)
with tarfile.open(archive) as tar:
names = tar.getnames()
index_name = names[0]
for x in FILES_TO_ARCHIVE:
assert os.path.join(index_name, x) in names, "File '{}' is missing in the archive".format(x)
index_dir = os.path.join(output_dir, index_name)
pro.message("Decompressing index core files")
cmd = ["tar", "xvf", archive, "-C", output_dir]
pro.run_safe(cmd)
pro.message("Core files have been decompressed, reconstructing the index")
pro.touch(os.path.join(index_dir, "index.fa"))
pro.touch(os.path.join(index_dir, "index.fa.pac"))
if klcp:
config = pro.load_index_config(index_dir)
cmd = [PROPHYLE, "index", "-k", config['k'], os.path.join(index_dir, "tree.nw"), index_dir]
else:
cmd = [PROPHYLE, "index", "-K", os.path.join(index_dir, "tree.nw"), index_dir]
pro.run_safe(cmd)
pro.message("Index reconstruction finished")
def _propagation_postprocessing(index_dir, in_tree_fn, out_tree_fn):
"""Merge reduced FASTA files after k-mer propagation and create index.fa.
Args:
index_dir (str): Index directory.
in_tree_fn (str): Input tree in Newick/NHX.
out_tree_fn (str): Output tree in Newick/NHX.
"""
pro.message('Propagation post-processing')
propagation_dir = os.path.join(index_dir, 'propagation')
tsv_fn = os.path.join(index_dir, "index.fa.kmers.tsv")
index_fa = os.path.join(index_dir, "index.fa")
command = ["cat", os.path.join(propagation_dir, "*.tsv"), '>', tsv_fn]
pro.run_safe(
command,
err_msg="K-mer statistics could not be created.",
thr_exc=True,
)
command = [PROPAGATION_POSTPROCESSING, propagation_dir, index_fa, in_tree_fn, tsv_fn, out_tree_fn]
pro.run_safe(
command,
err_msg="Main ProPhyle FASTA file could not be generated",
thr_exc=True,
)
pro.touch(index_fa + ".complete")
_log_file_md5(index_fa)
_log_file_md5(in_tree_fn)
_log_file_md5(out_tree_fn)
def _mark_complete(d, i=1, name=None):
"""Create a mark file (an empty file to mark a finished step nb i).
Args:
d (str): Directory.
i (int): Number of the step.
name (str): Name of the mark.
"""
assert i > 0
pro.touch(__mark_fn(d, i, name))
def prophyle_decompress(archive, output_dir, klcp):
pro.test_files(archive)
if not os.path.isdir(output_dir):
pro.error("Directory '{}' does not exist.".format(output_dir))
_compile_prophyle_bin(parallel=True)
with tarfile.open(archive) as tar:
names = tar.getnames()
index_name = names[0]
for x in FILES_TO_ARCHIVE:
if not os.path.join(index_name, x) in names:
pro.error("File '{}' is missing in the archive".format(x))
index_dir = os.path.join(output_dir, index_name)
index_exists = True
for i in range(1, 7):
fn = os.path.join(index_dir, ".complete.{}".format(i))
if not os.path.isfile(fn):
index_exists = False
break
if index_exists:
pro.message("Index already exists")
return
_compile_prophyle_bin(parallel=True)
pro.message("Decompressing core index files")
cmd = ["tar", "xvf", archive, "-C", output_dir]
pro.run_safe(cmd)
fn = os.path.join(index_dir, ".complete.4")
pro.rm(fn)
pro.message("Reconstructing the index")
pro.touch(os.path.join(index_dir, "index.fa"))
pro.touch(os.path.join(index_dir, "index.fa.pac"))
if klcp:
config = pro.load_index_config(index_dir)
cmd = [
PROPHYLE, "index", "-k", config['k'],
os.path.join(index_dir, "tree.nw"), index_dir
]
else:
cmd = [
PROPHYLE, "index", "-K",
os.path.join(index_dir, "tree.nw"), index_dir
]
pro.run_safe(cmd)
pro.message("Index reconstruction finished")