def __init__(self, source, dest=None):
# Basic #
self.source = FASTQ(source)
self.dest = DirectoryPath(dest)
# Default case #
if dest is None:
self.dest = DirectoryPath(self.source.prefix_path + '.fastqc')
def __init__(self, source, dest=None):
# Basic #
self.source = FASTQ(source)
self.dest = DirectoryPath(dest)
# Default case #
if dest is None:
self.dest = DirectoryPath(self.source.prefix_path + '.fastqc')
def __init__(self, path):
# Super #
DirectoryPath.__init__(self, path)
# The git directory #
self.git_dir = self.path + '.git'
# Check exists #
if not os.path.exists(self.git_dir):
raise Exception("No git repository at '%s'" % (self.git_dir))
def __init__(self, path):
# Super #
DirectoryPath.__init__(self, path)
# The git directory #
self.git_dir = self.path + '.git'
# Check exists #
if not os.path.exists(self.git_dir):
raise Exception("No git repository at '%s'" % (self.git_dir))
def __init__(self, path, empty=False):
# Super #
DirectoryPath.__init__(self, path)
# The git directory #
self.git_dir = self.path + '.git'
# Check exists #
if not empty and not self:
raise Exception("No git repository at '%s'" % (self.git_dir))
# Default arguments #
self.default = ["--git-dir=" + self.git_dir, "--work-tree=" + self.path]
def __init__(self,
command,
language='python',
base_dir=os.path.abspath(os.getcwd()),
modules=None,
**kwargs):
# Check the modules variable is a list #
if modules is None: self.modules = []
elif not isinstance(modules, list): self.modules = list(modules)
else: self.modules = modules
# Check command type #
if not isinstance(command, list): command = [command]
# Log directory #
for i in range(30):
now = datetime.datetime.now(dateutil.tz.tzlocal())
log_name = now.strftime("%Y-%m-%da%Hh%Mm%Ss%Z%z")
base_dir = DirectoryPath(base_dir + log_name + '/')
if not base_dir.exists:
base_dir.create()
break
else:
time.sleep(2)
continue
else:
base_dir.create()
# Modules directory #
modules_dir = DirectoryPath(base_dir + "modules/")
modules_dir.create()
# The script to be sent #
script = []
# Copy modules to the log directory #
for module in self.modules:
module_dir = os.path.dirname(module.__file__)
module_name = module.__name__
repos_dir = os.path.abspath(module_dir + '/../')
project_name = os.path.basename(repos_dir)
static_module_dir = modules_dir + project_name + '/'
module_version = module.__version__ + ' ' + get_git_tag(repos_dir)
# Copy #
print "Making static copy of module '%s' for SLURM job..." % module_name
sh.cp('-R', repos_dir, static_module_dir)
# Make script #
script.insert(0, "sys.path.insert(0, '%s')" % static_module_dir)
script += ["import %s" % module_name]
script += [
"print 'Using static copy of module %s version %s'" %
(module_name, module_version)
]
# Prepend to the script to be sent #
script.insert(0, "import os, sys")
# Add the user's command to the script #
script += command
# Super #
JobSLURM.__init__(self, script, language, base_dir, **kwargs)
def make_raw_zip_bundle(cluster):
"""Zip all the raw samples files in one file"""
result_dir = DirectoryPath(cluster.base_dir + "raw_reads")
result_dir.remove()
result_dir.create()
for sample in tqdm(cluster.samples):
subdir = DirectoryPath(result_dir + sample.short_name)
subdir.create()
shutil.copy(sample.raw.fwd, subdir)
shutil.copy(sample.raw.rev, subdir)
result_dir.zip(keep_orig=False)
def set_paths(self, base_dir, script_path):
"""Set the directory, the script path and the outfile path"""
# Make absolute paths #
if 'change_dir' in self.kwargs:
self.kwargs['change_dir'] = DirectoryPath(
os.path.abspath(self.kwargs['change_dir']))
if 'out_file' in self.kwargs:
self.kwargs['out_file'] = FilePath(
os.path.abspath(self.kwargs['out_file']))
# In case there is a base directory #
if base_dir is not None:
self.base_dir = DirectoryPath(os.path.abspath(base_dir))
self.script_path = FilePath(base_dir + "run." +
self.extensions[self.language])
self.kwargs['change_dir'] = base_dir
self.kwargs['out_file'] = FilePath(base_dir + "run.out")
# Other cases #
if base_dir is None and script_path is None:
self.script_path = FilePath(new_temp_path())
if script_path is not None:
self.script_path = FilePath(os.path.abspath(script_path))
def __init__(self, command,
language = 'python',
base_dir = os.path.abspath(os.getcwd()),
modules = None,
**kwargs):
# Check the modules variable is a list #
if modules is None: self.modules = []
elif not isinstance(modules, list): self.modules = list(modules)
else: self.modules = modules
# Check command type #
if not isinstance(command, list): command = [command]
# Log directory #
for i in range(30):
now = datetime.datetime.now(dateutil.tz.tzlocal())
log_name = now.strftime("%Y-%m-%da%Hh%Mm%Ss%Z%z")
base_dir = DirectoryPath(base_dir + log_name + '/')
if not base_dir.exists:
base_dir.create()
break
else:
time.sleep(2)
continue
else: base_dir.create()
# Modules directory #
modules_dir = DirectoryPath(base_dir + "modules/")
modules_dir.create()
# The script to be sent #
script = []
# Copy modules to the log directory #
for module in self.modules:
module_dir = os.path.dirname(module.__file__)
module_name = module.__name__
repos_dir = GitRepo(os.path.abspath(module_dir + '/../'))
project_name = os.path.basename(repos_dir)
static_module_dir = modules_dir + project_name + '/'
module_version = module.__version__ + ' ' + repos_dir.tag
# Copy #
print "Making static copy of module '%s' for SLURM job..." % module_name
sh.cp('-R', repos_dir, static_module_dir)
# Make script #
script.insert(0, "sys.path.insert(0, '%s')" % static_module_dir)
script += ["import %s" % module_name]
script += ["print 'Using static copy of module %s version %s'" % (module_name, module_version)]
# Prepend to the script to be sent #
script.insert(0, "import os, sys")
# Add the user's command to the script #
script += command
# Super #
JobSLURM.__init__(self, script, language, base_dir, **kwargs)
class FastQC(object):
"""Takes care of running the FastQC program.
See http://www.bioinformatics.babraham.ac.uk/projects/fastqc/
Expects version 0.10.1."""
def __init__(self, source, dest=None):
# Basic #
self.source = FASTQ(source)
self.dest = DirectoryPath(dest)
# Default case #
if dest is None:
self.dest = DirectoryPath(self.source.prefix_path + '.fastqc')
def check(self):
assert sh.fastqc('--v', )
def run(self):
if self.dest is None:
sh.fastqc(self.source, '-q')
os.remove(self.source.prefix_path + '_fastqc.zip')
if self.dest is not None:
if self.dest.exists: self.dest.remove()
self.tmp_dir = new_temp_dir()
sh.fastqc(self.source, '-q', '-o', self.tmp_dir)
created_dir = self.tmp_dir + self.source.prefix.split('.')[0] + '_fastqc/'
shutil.move(created_dir, self.dest)
self.tmp_dir.remove()
return self.results
@property
def output_dir(self):
if self.dest is None: return self.source.split('.')[0] + '_fastqc/'
else: return self.dest
@property_cached
def results(self):
results = FastQCResults(self.output_dir)
if not results: self.run()
return results
