def main(self, args):
"""
Internal: provides mock bcl2fastq2 functionality
"""
# Build generic header
header = """BCL to FASTQ file converter
bcl2fastq v2.17.1.14
Copyright (c) 2007-2015 Illumina, Inc.
2015-12-17 14:08:00 [7fa113f3f780] Command-line invocation: bcl2fastq %s""" \
% ' '.join(args)
# Handle version request
if "--version" in args:
print header
return self._exit_code
# Deal with arguments
p = argparse.ArgumentParser()
p.add_argument("--runfolder-dir", action="store")
p.add_argument("--output-dir", action="store")
p.add_argument("--sample-sheet", action="store")
p.add_argument("--use-bases-mask", action="store")
p.add_argument("--barcode-mismatches", action="store")
p.add_argument("--minimum-trimmed-read-length", action="store")
p.add_argument("--mask-short-adapter-reads", action="store")
p.add_argument("--ignore-missing-bcls", action="store_true")
p.add_argument("--no-lane-splitting", action="store_true")
p.add_argument("-r", action="store")
p.add_argument("-d", action="store")
p.add_argument("-p", action="store")
p.add_argument("-w", action="store")
args = p.parse_args(args)
# Check bases mask
if self._assert_bases_mask:
print "Checking bases mask: %s" % args.use_bases_mask
assert (args.use_bases_mask == self._assert_bases_mask)
# Platform
print "Platform (default): %s" % self._platform
# Run folder (input data)
runfolder = args.runfolder_dir
print "Runfolder dir: %s" % runfolder
if runfolder is None:
return 1
run_info_xml = os.path.join(runfolder, "RunInfo.xml")
if not os.path.exists(run_info_xml):
return 1
# Determine if run is paired end
nreads = 0
for r in IlluminaRunInfo(run_info_xml).reads:
if r['is_indexed_read'] == 'N':
nreads += 1
if nreads == 2:
paired_end = True
else:
paired_end = False
print "Paired-end: %s" % paired_end
# Lanes
lanes = IlluminaRun(runfolder, platform=self._platform).lanes
print "Lanes: %s" % lanes
# Output folder
output_dir = args.output_dir
if output_dir is None:
output_dir = "bcl2fastq"
print "Output dir: %s" % output_dir
# Sample sheet
sample_sheet = args.sample_sheet
if sample_sheet is None:
for d in (runfolder,
os.path.join(runfolder, "Data", "Intensities",
"BaseCalls")):
sample_sheet = os.path.join(d, "SampleSheet.csv")
if os.path.exists(sample_sheet):
break
sample_sheet = None
print "Sample sheet: %s" % sample_sheet
# Modifiers
no_lane_splitting = bool(args.no_lane_splitting)
print "No lane splitting: %s" % no_lane_splitting
# Generate mock output based on inputs
tmpname = "tmp.%s" % uuid.uuid4()
output = MockIlluminaData(name=tmpname,
package="bcl2fastq2",
unaligned_dir="bcl2fastq")
missing_fastqs = self._missing_fastqs
# Add outputs from sample sheet (if supplied)
if sample_sheet is not None:
s = SampleSheetPredictor(sample_sheet_file=sample_sheet)
s.set(paired_end=paired_end,
no_lane_splitting=no_lane_splitting,
lanes=lanes)
for project in s.projects:
print "Adding project: %s" % project.name
for sample in project.samples:
for fq in sample.fastqs():
if missing_fastqs and (fq in missing_fastqs):
continue
if sample.sample_name is None:
sample_name = sample.sample_id
else:
sample_name = sample.sample_name
output.add_fastq(project.name, sample_name, fq)
# Add undetermined fastqs
# NB Would like to use the 'add_undetermined'
# method but this doesn't play well with using
# the predictor-based approach above
if paired_end:
reads = (1, 2)
else:
reads = (1, )
if no_lane_splitting:
lanes = None
for r in reads:
if lanes is None:
output.add_fastq("Undetermined_indices", "undetermined",
"Undetermined_S0_R%d_001.fastq.gz" % r)
else:
for lane in lanes:
output.add_fastq(
"Undetermined_indices", "undetermined",
"Undetermined_S0_L%03d_R%d_001.fastq.gz" % (lane, r))
# Build the output directory
output.create()
# Move to final location
os.rename(os.path.join(tmpname, "bcl2fastq"), output_dir)
shutil.rmtree(tmpname)
return self._exit_code
