undetermined_dir = os.path.abspath(args.undetermined_dir)
# Set up barcode data
barcodes = []
for barcode_info in args.barcode_info:
name,barcode,lane = barcode_info.split(':')
print "Assigning barcode '%s' in lane %s to %s" % (barcode,lane,name)
barcodes.append({ 'name': name,
'index': barcode,
'matcher': BarcodeMatcher(barcode),
'lane': int(lane)})
# Read from sample sheet (if supplied)
if args.sample_sheet is not None:
print "Reading data from sample sheet %s" % args.sample_sheet
sample_sheet = IlluminaData.CasavaSampleSheet(args.sample_sheet)
for line in sample_sheet:
name = line['SampleID']
barcode = line['Index'].rstrip('N').rstrip('-').rstrip('N')
lane = line['Lane']
print "Assigning barcode '%s' in lane %s to %s" % (barcode,lane,name)
barcodes.append({ 'name': name,
'index': barcode,
'matcher': BarcodeMatcher(barcode),
'lane': int(lane) })
if len(barcodes) < 1:
p.error("need at least one --barcode and/or --samplesheet assignment")
# Collect input files
p = IlluminaData.IlluminaProject(undetermined_dir)
logging.warning(
"--miseq option no longer necessary; MiSEQ-style sample sheets "
"are now converted automatically")
# Get input sample sheet file
samplesheet = args[0]
if not os.path.isfile(samplesheet):
logging.error("sample sheet '%s': not found" % samplesheet)
sys.exit(1)
# Read in the data as CSV
data = IlluminaData.get_casava_sample_sheet(samplesheet)
# Remove lanes
if options.lanes is not None:
lanes = parse_lane_expression(options.lanes)
print "Keeping lanes %s, removing the rest" % ','.join(
[str(x) for x in lanes])
new_data = IlluminaData.CasavaSampleSheet()
for line in data:
if line['Lane'] in lanes:
print "Keeping %s" % line
new_data.append(tabdata="%s" % line)
data = new_data
# Update the SampleID and SampleProject fields
for sample_id in options.sample_id:
lanes, name = parse_name_expression(sample_id)
for line in data:
if line['Lane'] in lanes:
print "Setting SampleID for lane %d: '%s'" % (line['Lane'],
name)
line['SampleID'] = name
# Update the SampleProject field
for sample_project in options.sample_project:
# Set up barcode data
barcodes = []
for barcode_info in options.barcode_info:
name, barcode, lane = barcode_info.split(':')
print "Assigning barcode '%s' in lane %s to %s" % (barcode, lane, name)
barcodes.append({
'name': name,
'index': barcode,
'matcher': BarcodeMatcher(barcode),
'lane': int(lane)
})
# Read from sample sheet (if supplied)
if options.sample_sheet is not None:
print "Reading data from sample sheet %s" % options.sample_sheet
sample_sheet = IlluminaData.CasavaSampleSheet(options.sample_sheet)
for line in sample_sheet:
name = line['SampleID']
barcode = line['Index'].rstrip('N').rstrip('-').rstrip('N')
lane = line['Lane']
print "Assigning barcode '%s' in lane %s to %s" % (barcode, lane,
name)
barcodes.append({
'name': name,
'index': barcode,
'matcher': BarcodeMatcher(barcode),
'lane': int(lane)
})
if len(barcodes) < 1:
p.error("need at least one --barcode and/or --samplesheet assignment")
