def _sample_based_run_info(fc):
"""Convert a regular run info file into a sample-based info file, one sample per lane"""
sample_fc = PostProcessedFlowcell(fc.get_fc_name(),
fc.get_fc_date(), {},
fc_alias=fc.get_fc_alias(),
fc_dir=fc.get_fc_dir())
lane_num = 0
for l in fc.get_lanes():
bcids = l.get_barcode_ids()
for barcode_id in bcids:
s = l.get_sample_by_barcode(barcode_id)
lane_num = lane_num + 1
newl = Lane(
data={
"description": str(s.get_name()),
"lane": lane_num,
"multiplex": [],
"analysis": "Minimal",
"genome_build": s.get_genome_build(),
"sample_prj": s.get_project()
})
newl.set_name("%s" % lane_num)
files = l.get_files()
if options.customer_delivery or options.barcode_id_to_name:
pat = "%s.*_%s_[12].*$" % (
l.get_name(), get_sample_name(s.get_barcode_name()))
else:
pat = "%s.*_%s_[12].*$" % (l.get_name(), barcode_id)
bc_files = re.compile(pat, re.IGNORECASE)
files = filter(bc_files.search, files)
newl.set_files(files)
sample_fc.add_lane(newl)
_save_run_info(sample_fc, "sample_project_run_info.yaml")
def _convert_barcode_id_to_name(multiplex, fc_name, fq):
bcid2name = dict([(str(mp.get_barcode_id()), mp.get_full_name()) for mp in multiplex])
if options.no_full_names:
bcid2name = dict([(str(mp.get_barcode_id()), get_sample_name(mp.get_barcode_name())) for mp in multiplex])
bcid = re.search("_(\d+)_(\d+)_fastq.txt", fq)
from_str = "%s_%s_fastq.txt" % (bcid.group(1), bcid.group(2))
to_str = "%s_%s.fastq" % (bcid2name[bcid.group(1)], bcid.group(2))
return fq.replace(from_str, to_str)
def _convert_barcode_id_to_name(multiplex, fc_name, fq):
bcid2name = dict([(str(mp.get_barcode_id()), mp.get_full_name())
for mp in multiplex])
if options.no_full_names:
bcid2name = dict([(str(mp.get_barcode_id()),
get_sample_name(mp.get_barcode_name()))
for mp in multiplex])
bcid = re.search("_(\d+)_(\d+)_fastq.txt", fq)
from_str = "%s_%s_fastq.txt" % (bcid.group(1), bcid.group(2))
to_str = "%s_%s.fastq" % (bcid2name[bcid.group(1)], bcid.group(2))
return fq.replace(from_str, to_str)
def _sample_based_run_info(fc):
"""Convert a regular run info file into a sample-based info file, one sample per lane"""
sample_fc = PostProcessedFlowcell(fc.get_fc_name(), fc.get_fc_date(), {}, fc_alias = fc.get_fc_alias(), fc_dir=fc.get_fc_dir())
lane_num = 0
for l in fc.get_lanes():
bcids = l.get_barcode_ids()
for barcode_id in bcids:
s = l.get_sample_by_barcode(barcode_id)
lane_num = lane_num + 1
newl = Lane(data={"description":str(s.get_name()), "lane" :lane_num, "multiplex":[], "analysis":"Minimal", "genome_build":s.get_genome_build(), "sample_prj":s.get_project()})
newl.set_name("%s" % lane_num)
files = l.get_files()
if options.customer_delivery or options.barcode_id_to_name:
pat = "%s.*_%s_[12].*$" % (l.get_name(), get_sample_name(s.get_barcode_name()))
else:
pat = "%s.*_%s_[12].*$" % (l.get_name(), barcode_id)
bc_files = re.compile(pat, re.IGNORECASE)
files = filter(bc_files.search, files)
newl.set_files(files)
sample_fc.add_lane(newl)
_save_run_info(sample_fc, "sample_project_run_info.yaml")