def _build_samples_from_dir(data_fp, filename_fmt, excluded):
"""
Build a list of samples from a data filepath and filename format.
:param data_fp: a Path to data files
:param filename_fmt: a string giving wildcards for {sample} and (optionally)
{rp}, for read pair (e.g. R1 or R2).
:param exclude: a list of sample names to exclude
:returns: A dictionary of samples, with sample names as keys
"""
files = list(listfiles(str(data_fp / filename_fmt)))
Samples = {
t[1]['sample']: {}
for t in files if t[1]['sample'] not in excluded
}
for f in files:
fpath = f[0]
wcards = f[1]
if wcards['sample'] in excluded:
continue
rp = wcards['rp'] if 'rp' in wcards.keys() else False
if rp:
Samples[wcards['sample']][rp] = fpath
Samples[wcards['sample']]['paired'] = True
else:
Samples[wcards['sample']]['file'] = fpath
Samples[wcards['sample']]['paired'] = False
return Samples
def find_samples(data_fp, filename_fmt):
"""
Build a list of samples from a data filepath and filename format.
:param data_fp: a Path to data files
:param filename_fmt: a string giving wildcards for {sample} and (optionally)
{rp}, for read pair (e.g. R1 or R2).
:returns: A dictionary of samples, with sample names as keys:
Samples = {
'sample1': {
'1': 'path/to/sample1_R1.fastq.gz',
'2': 'path/to/sample1_R2.fastq.gz', #optional
}, ...
}
"""
files = list(listfiles(str(data_fp / filename_fmt)))
Samples = {t[1]['sample']: {} for t in files}
for f in files:
fpath = f[0]
wcards = f[1]
rp = wcards.get('rp')
if rp:
if not rp in ['1', '2']:
raise ValueError(
"'{rp}' should capture just '1' or '2' in filename, nothing else"
)
Samples[wcards['sample']][rp] = fpath
else:
Samples[wcards['sample']]['1'] = fpath
return Samples
def _locks(self, type):
return (
f
for f, _ in listfiles(
os.path.join(self._lockdir, "{{n,[0-9]+}}.{}.lock".format(type))
)
if not os.path.isdir(f)
)
def clear_generated_files(include_targets=False):
cleared_files = 0
for pattern in all_intermediate_files:
from snakemake.utils import listfiles
for filename, wildcards in listfiles(pattern):
if os.path.isfile(filename):
os.unlink(filename)
cleared_files += 1
else:
if cleared_files > 0:
print('{} intermediate file(s) were removed.'.format(cleared_files))
for dir in INTERMEDIATE_DIRS:
if os.path.islink(dir):
realpath = os.readlink(dir)
print('Cleaning a symbolic link:', dir)
shutil.rmtree(realpath, ignore_errors=True)
os.unlink(dir)
elif os.path.isdir(dir):
print('Cleaning an intermediate directory:', dir)
shutil.rmtree(dir, ignore_errors=True)
if not include_targets:
return
from itertools import groupby
files_safe_to_delete = {'.AppleDouble', '.DS_Store'}
for dirname, files in groupby(sorted(TARGETS), lambda x: x.split(os.sep)[0]):
for f in files:
if os.path.exists(f):
os.unlink(f)
if os.path.isdir(dirname) and not (
set(os.listdir(dirname)) - files_safe_to_delete):
shutil.rmtree(dirname)
def expand_dynamic(self, pattern):
""" Expand dynamic files. """
return list(
listfiles(pattern,
restriction=self.wildcards,
omit_value=DYNAMIC_FILL))
def expand_dynamic(pattern, restriction=None, omit_value=None):
""" Expand dynamic files. """
return list(
listfiles(pattern, restriction=restriction, omit_value=omit_value))
def expand_dynamic(self, pattern):
""" Expand dynamic files. """
return list(listfiles(pattern,
restriction=self.wildcards,
omit_value=DYNAMIC_FILL))
def expand_dynamic(pattern, restriction=None, omit_value=None):
""" Expand dynamic files. """
return list(listfiles(pattern,
restriction=restriction,
omit_value=omit_value))
def _locks(self, type):
return (f for f, _ in listfiles(
os.path.join(self._lockdir, "{{n,[0-9]+}}.{}.lock".format(type)))
if not os.path.isdir(f))