def get_library_names_and_timestamps(name):
"""
Extract sample/library names and timestamps
Given a name from 'get_generic_name', extract the sample,
library and timestamp elements.
If no timestamp is present then it is returned
as the string 'unknown'.
For 'standard' SOLiD data directories, the top-level is a
'sample', which may be split into multiple 'libraries', and
then again in multiple timestamped filesets.
Some examples:
* LH_POOL/results.F1B1/libraries/LH1/primary.20111208144829752/reads/solid0127_20111207_FRAG_BC_LH_POOL_BC_LH1
* ZD_hu/results.F1B1/primary.20091220022109452/reads/solid0424_20091214_ZD_hu_F3
* SH_JC1_pool/results.F1B1/libraries_MM2/JC_SEQ30/primary.20120125063517232/reads/solid0127_20120117_PE_BC_SH_JC1_pool_F5-BC_JC_SEQ30
"""
# Extract timestamp
timestamp = SolidData.extract_library_timestamp(name)
if timestamp is None:
timestamp = 'unknown'
# Attempt to extract sample and library from the path
new_path = []
for field in name.split(os.sep)[0:-1]:
if field == "results.F1B1":
continue
elif field.startswith("libraries"):
continue
elif field == "reads":
continue
elif field.startswith("primary.") or field.startswith("secondary."):
continue
new_path.append(field)
if len(new_path) < 2:
sample = new_path[0]
library = None
else:
sample,library = new_path[0:2]
# Return
return (sample,library,timestamp)
def get_library_names_and_timestamps(name):
"""
Extract sample/library names and timestamps
Given a name from 'get_generic_name', extract the sample,
library and timestamp elements.
If no timestamp is present then it is returned
as the string 'unknown'.
For 'standard' SOLiD data directories, the top-level is a
'sample', which may be split into multiple 'libraries', and
then again in multiple timestamped filesets.
Some examples:
* LH_POOL/results.F1B1/libraries/LH1/primary.20111208144829752/reads/solid0127_20111207_FRAG_BC_LH_POOL_BC_LH1
* ZD_hu/results.F1B1/primary.20091220022109452/reads/solid0424_20091214_ZD_hu_F3
* SH_JC1_pool/results.F1B1/libraries_MM2/JC_SEQ30/primary.20120125063517232/reads/solid0127_20120117_PE_BC_SH_JC1_pool_F5-BC_JC_SEQ30
"""
# Extract timestamp
timestamp = SolidData.extract_library_timestamp(name)
if timestamp is None:
timestamp = 'unknown'
# Attempt to extract sample and library from the path
new_path = []
for field in name.split(os.sep)[0:-1]:
if field == "results.F1B1":
continue
elif field.startswith("libraries"):
continue
elif field == "reads":
continue
elif field.startswith("primary.") or field.startswith("secondary."):
continue
new_path.append(field)
if len(new_path) < 2:
sample = new_path[0]
library = None
else:
sample, library = new_path[0:2]
# Return
return (sample, library, timestamp)
def verify_runs(solid_dirs):
"""Do basic verification checks on SOLiD run directories
For each SOLiD run directory, create a SolidRun object and check for the
expected sample and library directories, and that primary data files
(csfasta and qual) have been assigned and exist.
Returns a UNIX-like status code: 0 indicates that the checks passed,
1 indicates that they failed.
Arguments:
solid_dirs: a list of SOLiD sequencing directory names.
Returns:
0 if the run is verified, 1 if there is a problem.
"""
print "Performing verification"
status = 0
for solid_dir in solid_dirs:
# Initialise
run_status = 0
run = SolidData.SolidRun(solid_dir)
if not run.verify():
run_status = 1
print "%s:" % run.run_name,
if run_status == 0:
print " [PASSED]"
else:
print " [FAILED]"
status = 1
# Completed
print "Overall status:",
if status == 0:
print " [PASSED]"
else:
print " [FAILED]"
return status
logging.getLogger().setLevel(logging.DEBUG)
elif options.no_warnings:
logging.getLogger().setLevel(logging.ERROR)
# Solid run directories
for arg in args:
if not os.path.isdir(arg):
logging.error("'%s' not found or not a directory" % arg)
sys.exit(1)
if len(args) == 1:
# Single directory supplied
if options.only:
solid_dirs = [args[0]]
else:
# Add associated directories
solid_dirs = SolidData.list_run_directories(args[0])
else:
# Use all supplied arguments
solid_dirs = args
# Output spreadsheet name
if options.xls:
spreadsheet = os.path.splitext(os.path.basename(
solid_dirs[0]))[0] + ".xls"
print "Writing spreadsheet %s" % spreadsheet
# Check there's at least one thing to do
if not (options.report or options.layout or options.xls or options.verify
or options.rsync or options.md5sum or options.copy_pattern
or options.gzip_pattern or options.md5_pattern):
options.report = True
logging.getLogger().setLevel(logging.DEBUG)
elif options.no_warnings:
logging.getLogger().setLevel(logging.ERROR)
# Solid run directories
for arg in args:
if not os.path.isdir(arg):
logging.error("'%s' not found or not a directory" % arg)
sys.exit(1)
if len(args) == 1:
# Single directory supplied
if options.only:
solid_dirs = [args[0]]
else:
# Add associated directories
solid_dirs = SolidData.list_run_directories(args[0])
else:
# Use all supplied arguments
solid_dirs = args
# Output spreadsheet name
if options.xls:
spreadsheet = os.path.splitext(os.path.basename(solid_dirs[0]))[0] + ".xls"
print "Writing spreadsheet %s" % spreadsheet
# Check there's at least one thing to do
if not (options.report or
options.layout or
options.xls or
options.verify or
options.rsync or