def verify_fastq_generation(ap,
unaligned_dir=None,
lanes=None,
include_sample_dir=False):
"""Check that generated Fastqs match sample sheet predictions
Arguments:
ap (AutoProcessor): autoprocessor pointing to the analysis
directory to do Fastqs verification on
unaligned_dir (str): explicitly specify the bcl2fastq output
directory to check
lanes (list): specify a list of lane numbers (integers) to
check (others will be ignored)
include_sample_dir (bool): if True then include a
'sample_name' directory level when checking for
bcl2fastq2 outputs, even if one shouldn't be present
Returns:
True if outputs match sample sheet, False otherwise.
"""
if unaligned_dir is None:
if ap.params.unaligned_dir is not None:
unaligned_dir = ap.params.unaligned_dir
else:
raise Exception("Bcl2fastq output directory not defined")
print "Checking bcl2fastq output directory '%s'" % unaligned_dir
bcl_to_fastq_dir = os.path.join(ap.analysis_dir, unaligned_dir)
if not os.path.isdir(bcl_to_fastq_dir):
# Directory doesn't exist
return False
# Make a temporary sample sheet to verify against
tmp_sample_sheet = os.path.join(
ap.tmp_dir,
"SampleSheet.verify.%s.csv" % time.strftime("%Y%m%d%H%M%S"))
make_custom_sample_sheet(ap.params.sample_sheet,
tmp_sample_sheet,
lanes=lanes)
# Try to create an IlluminaData object
try:
illumina_data = IlluminaData.IlluminaData(ap.analysis_dir,
unaligned_dir=unaligned_dir)
except IlluminaData.IlluminaDataError as ex:
# Failed to initialise
logger.warning("Failed to get information from %s: %s" %
(bcl_to_fastq_dir, ex))
return False
# Do check
return IlluminaData.verify_run_against_sample_sheet(
illumina_data, tmp_sample_sheet, include_sample_dir=include_sample_dir)
for sample in project.samples:
if bcf_utils.name_matches(sample.name, sample_pattern):
for fastq in sample.fastq:
fastq_file = os.path.join(sample.dirn, fastq)
print "\tCopying .../%s" % os.path.basename(
fastq_file)
dst = os.path.abspath(os.path.basename(fastq_file))
if os.path.exists(dst):
logging.error(
"File %s already exists! Skipped" % dst)
else:
shutil.copy(fastq_file, dst)
# Verify against sample sheet
if options.sample_sheet is not None:
if IlluminaData.verify_run_against_sample_sheet(
illumina_data, options.sample_sheet):
print "Verification against sample sheet '%s': OK" % \
options.sample_sheet
status = 0
else:
logging.error("Verification against sample sheet '%s': FAILED" %
options.sample_sheet)
status = 1
sys.exit(status)
# Merge multiple fastqs in each sample
if options.merge_fastqs:
for project in illumina_data.projects:
for sample in project.samples:
for read in (1, 2):
# Concatenate fastqs for this read
if bcf_utils.name_matches(project.name,project_pattern):
# Loop through samples
for sample in project.samples:
if bcf_utils.name_matches(sample.name,sample_pattern):
for fastq in sample.fastq:
fastq_file = os.path.join(sample.dirn,fastq)
print "\tCopying .../%s" % os.path.basename(fastq_file)
dst = os.path.abspath(os.path.basename(fastq_file))
if os.path.exists(dst):
logging.error("File %s already exists! Skipped" % dst)
else:
shutil.copy(fastq_file,dst)
# Verify against sample sheet
if options.sample_sheet is not None:
if IlluminaData.verify_run_against_sample_sheet(illumina_data,options.sample_sheet):
print "Verification against sample sheet '%s': OK" % \
options.sample_sheet
status = 0
else:
logging.error("Verification against sample sheet '%s': FAILED" %
options.sample_sheet)
status = 1
sys.exit(status)
# Merge multiple fastqs in each sample
if options.merge_fastqs:
for project in illumina_data.projects:
for sample in project.samples:
for read in (1,2):
# Concatenate fastqs for this read
