def _run_qc_fastqc(in_files, data, out_dir):
in_files = fastq.downsample(in_files[0], in_files[1], N=5000000)
for fastq_file in in_files:
if fastq_file:
fastqc.run(
fastq_file, data,
os.path.join(
out_dir,
utils.splitext_plus(os.path.basename(fastq_file))[0]))
def _fastqc(input_file, out_dir):
data = {'config': {'algorithm': {}}}
if not file_exists(out_dir):
dw_file, _ = fastq.downsample(input_file, None, data, int(1e7))
cmd = ("fastqc {dw_file} --extract -o {out_dir}")
out_dir = os.path.abspath(out_dir)
safe_makedir(out_dir)
do.run(cmd.format(**locals()), "Doing Fastqc %s" % input_file)
logger.my_logger.debug(cmd.format(**locals()))
return out_dir
def process_lane(item):
"""Prepare lanes, potentially splitting based on barcodes and reducing the
number of reads for a test run
"""
NUM_DOWNSAMPLE = 10000
logger.debug("Preparing %s" % item["rgnames"]["lane"])
file1, file2 = get_fastq_files(item)
if item.get("test_run", False):
if bam.is_bam(file1):
file1 = bam.downsample(file1, item, NUM_DOWNSAMPLE)
else:
file1, file2 = fastq.downsample(file1, file2, item, NUM_DOWNSAMPLE, quick=True)
item["files"] = (file1, file2)
return [item]
def prepare_sample(data):
"""Prepare a sample to be run, potentially converting from BAM to
FASTQ and/or downsampling the number of reads for a test run
"""
NUM_DOWNSAMPLE = 10000
logger.debug("Preparing %s" % data["rgnames"]["sample"])
file1, file2 = get_fastq_files(data)
if data.get("test_run", False):
if bam.is_bam(file1):
file1 = bam.downsample(file1, data, NUM_DOWNSAMPLE)
file2 = None
else:
file1, file2 = fastq.downsample(file1, file2, data, NUM_DOWNSAMPLE, quick=True)
data["files"] = [file1, file2]
return [[data]]
def process_lane(item):
"""Prepare lanes, potentially splitting based on barcodes and reducing the
number of reads for a test run
"""
NUM_DOWNSAMPLE = 10000
logger.debug("Preparing %s" % item["rgnames"]["lane"])
file1, file2 = get_fastq_files(item)
if item.get("test_run", False):
if bam.is_bam(file1):
file1 = bam.downsample(file1, item, NUM_DOWNSAMPLE)
file2 = None
else:
file1, file2 = fastq.downsample(file1, file2, item,
NUM_DOWNSAMPLE, quick=True)
item["files"] = [file1, file2]
return [[item]]
def prepare_sample(data):
"""Prepare a sample to be run, potentially converting from BAM to
FASTQ and/or downsampling the number of reads for a test run
"""
NUM_DOWNSAMPLE = 10000
logger.debug("Preparing %s" % data["rgnames"]["sample"])
file1, file2 = get_fastq_files(data)
if data.get("test_run", False):
if bam.is_bam(file1):
file1 = bam.downsample(file1, data, NUM_DOWNSAMPLE)
file2 = None
else:
file1, file2 = fastq.downsample(file1,
file2,
data,
NUM_DOWNSAMPLE,
quick=True)
data["files"] = [file1, file2]
return [[data]]
