def test_make_fastqs_icell8_protocol(self):
"""make_fastqs: icell8 protocol
"""
# Create mock source data
illumina_run = MockIlluminaRun("171020_SN7001250_00002_AHGXXXX",
"hiseq",
top_dir=self.wd)
illumina_run.create()
# Create mock bcl2fastq
# Check that bases mask is as expected
MockBcl2fastq2Exe.create(os.path.join(self.bin, "bcl2fastq"),
assert_bases_mask="y25n76,I8,I8,y101")
os.environ['PATH'] = "%s:%s" % (self.bin, os.environ['PATH'])
# Do the test
ap = AutoProcess()
ap.setup(os.path.join(self.wd, "171020_SN7001250_00002_AHGXXXX"))
self.assertTrue(ap.params.sample_sheet is not None)
ap.make_fastqs(protocol="icell8")
# Check outputs
analysis_dir = os.path.join(self.wd,
"171020_SN7001250_00002_AHGXXXX_analysis")
for subdir in (os.path.join("primary_data",
"171020_SN7001250_00002_AHGXXXX"),
os.path.join("logs",
"002_make_fastqs_icell8"), "bcl2fastq"):
self.assertTrue(os.path.isdir(os.path.join(analysis_dir, subdir)),
"Missing subdir: %s" % subdir)
for filen in ("statistics.info", "statistics_full.info",
"per_lane_statistics.info", "per_lane_sample_stats.info",
"projects.info", "processing_qc.html"):
self.assertTrue(os.path.isfile(os.path.join(analysis_dir, filen)),
"Missing file: %s" % filen)
def test_make_fastqs_specify_platform_via_metadata(self):
"""make_fastqs: implicitly specify the platform via metadata
"""
# Create mock source data
illumina_run = MockIlluminaRun("171020_UNKNOWN_00002_AHGXXXX",
"miseq",
top_dir=self.wd)
illumina_run.create()
# Create mock bcl2fastq
MockBcl2fastq2Exe.create(os.path.join(self.bin, "bcl2fastq"),
platform="miseq")
os.environ['PATH'] = "%s:%s" % (self.bin, os.environ['PATH'])
# Do the test
ap = AutoProcess()
ap.setup(os.path.join(self.wd, "171020_UNKNOWN_00002_AHGXXXX"))
self.assertTrue(ap.params.sample_sheet is not None)
self.assertTrue(ap.metadata.platform is None)
ap.metadata["platform"] = "miseq"
ap.make_fastqs(protocol="standard")
# Check outputs
analysis_dir = os.path.join(self.wd,
"171020_UNKNOWN_00002_AHGXXXX_analysis")
for subdir in (os.path.join("primary_data",
"171020_UNKNOWN_00002_AHGXXXX"),
os.path.join("logs", "002_make_fastqs"), "bcl2fastq"):
self.assertTrue(os.path.isdir(os.path.join(analysis_dir, subdir)),
"Missing subdir: %s" % subdir)
for filen in ("statistics.info", "statistics_full.info",
"per_lane_statistics.info", "per_lane_sample_stats.info",
"projects.info", "processing_qc.html"):
self.assertTrue(os.path.isfile(os.path.join(analysis_dir, filen)),
"Missing file: %s" % filen)
def test_make_fastqs_missing_fastqs_with_placeholders(self):
"""make_fastqs: missing fastqs with placeholders
"""
# Create mock source data
illumina_run = MockIlluminaRun("171020_M00879_00002_AHGXXXX",
"miseq",
top_dir=self.wd)
illumina_run.create()
# Create mock bcl2fastq
MockBcl2fastq2Exe.create(os.path.join(self.bin, "bcl2fastq"),
missing_fastqs=(
"Sample1_S1_L001_R1_001.fastq.gz",
"Sample1_S1_L001_R2_001.fastq.gz",
))
os.environ['PATH'] = "%s:%s" % (self.bin, os.environ['PATH'])
# Do the test
ap = AutoProcess()
ap.setup(os.path.join(self.wd, "171020_M00879_00002_AHGXXXX"))
self.assertTrue(ap.params.sample_sheet is not None)
ap.make_fastqs(protocol="standard", create_empty_fastqs=True)
# Check outputs
analysis_dir = os.path.join(self.wd,
"171020_M00879_00002_AHGXXXX_analysis")
for subdir in (os.path.join("primary_data",
"171020_M00879_00002_AHGXXXX"),
os.path.join("logs", "002_make_fastqs"), "bcl2fastq"):
self.assertTrue(os.path.isdir(os.path.join(analysis_dir, subdir)),
"Missing subdir: %s" % subdir)
for filen in ("statistics.info", "statistics_full.info",
"per_lane_statistics.info", "per_lane_sample_stats.info",
"projects.info"):
self.assertTrue(os.path.exists(os.path.join(analysis_dir, filen)),
"Missing file: %s" % filen)
self.assertTrue(
os.path.exists(
os.path.join(analysis_dir, "logs", "002_make_fastqs",
"missing_fastqs.log")))
elif options.use_lane_splitting:
no_lane_splitting = False
else:
no_lane_splitting = None
# Do the make_fastqs step
d.make_fastqs(skip_rsync=options.skip_rsync,
nprocessors=options.nprocessors,
runner=options.runner,
remove_primary_data=options.remove_primary_data,
ignore_missing_bcl=options.ignore_missing_bcl,
ignore_missing_stats=options.ignore_missing_stats,
generate_stats=(not options.no_stats),
require_bcl2fastq_version=options.bcl2fastq_version,
unaligned_dir=options.unaligned_dir,
sample_sheet=options.sample_sheet,
bases_mask=options.bases_mask,
no_lane_splitting=no_lane_splitting,
minimum_trimmed_read_length=options.minimum_trimmed_read_length,
mask_short_adapter_reads=options.mask_short_adapter_reads,
stats_file=options.stats_file,
per_lane_stats_file=options.per_lane_stats_file,
report_barcodes=options.report_barcodes,
barcodes_file=options.barcodes_file,
skip_bcl2fastq=options.skip_bcl2fastq,
only_fetch_primary_data=options.only_fetch_primary_data,
create_empty_fastqs=options.create_empty_fastqs)
elif cmd == 'merge_fastq_dirs':
d.merge_fastq_dirs(options.unaligned_dir,
dry_run=options.dry_run)
elif cmd == 'update_fastq_stats':
d.set_log_dir(d.get_log_subdir(cmd))
