def run(self):
# Check #
if glob.glob('454Reads.*.sff'):
raise Exception(
"Demultiplexed files already present in current directory.")
# Extract #
for sff in self.sff_files:
sff.split()
# Assign #
for sample in self.samples:
sample.pieces = []
for sff in self.sff_files:
sff.assign_pieces()
# Report #
columns = [sff.name for sff in self.sff_files]
rows = [sample.name for sample in self.samples]
data = [[sff.count_reads_in(sample) for sff in self.sff_files]
for sample in self.samples]
self.frame = pandas.DataFrame(data, index=rows, columns=columns)
self.frame['Total'] = self.frame.sum(1)
# Regroup #
for sample in self.samples:
if len(sample.pieces) == 1:
shutil.move(sample.pieces[0].path, sample.p.raw_sff)
else:
sh.sfffile("-o", sample.p.raw_sff,
*[p.path for p in sample.pieces])
for p in sample.pieces:
os.remove(p.path)
# Save report #
self.frame.to_csv(illumitag.view_dir + 'pyrosamples/report.csv')
def split(self):
# Call Roche binary #
barcode_file = TmpFile.from_string(self.barcode_text)
sh.sfffile("-s", "barcodes_keyword", "-mcf", barcode_file.path, self.path)
# Check result #
produced_files = set(sh.glob('454Reads.*.sff'))
expected_files = set(['454Reads.%s.sff' % (sample.name.upper()) for sample in self.sample_links.values()])
assert produced_files == expected_files
# Make piece objects #
self.pieces = [SamplePiece(p, self) for p in sh.glob('454Reads.*.sff')]
for piece in self.pieces: piece.rename()
# Cleanup #
barcode_file.remove()
def split(self):
# Call Roche binary #
handle = tempfile.NamedTemporaryFile(delete=False)
handle.write(self.barcode_file)
handle.close()
sh.sfffile("-s", "barcodes_keyword", "-mcf", handle.name, self.path)
# Check result #
produced_files = set(sh.glob('454Reads.*.sff'))
expected_files = set(['454Reads.%s.sff' % (sample.name.upper()) for sample in self.sample_links.values()])
assert produced_files == expected_files
# Make piece objects #
self.pieces = [SamplePiece(p, self) for p in sh.glob('454Reads.*.sff')]
for piece in self.pieces: piece.rename()
os.remove(handle.name)
def split(self):
# Call Roche binary #
barcode_file = TmpFile.from_string(self.barcode_text)
sh.sfffile("-s", "barcodes_keyword", "-mcf", barcode_file.path,
self.path)
# Check result #
produced_files = set(sh.glob('454Reads.*.sff'))
expected_files = set([
'454Reads.%s.sff' % (sample.name.upper())
for sample in self.sample_links.values()
])
assert produced_files == expected_files
# Make piece objects #
self.pieces = [SamplePiece(p, self) for p in sh.glob('454Reads.*.sff')]
for piece in self.pieces:
piece.rename()
# Cleanup #
barcode_file.remove()
def run(self):
# Check #
if glob.glob('454Reads.*.sff'): raise Exception("Demultiplexed files already present in current directory.")
# Extract #
for sff in self.sff_files: sff.split()
# Assign #
for sample in self.samples: sample.pieces = []
for sff in self.sff_files: sff.assign_pieces()
# Report #
columns = [sff.name for sff in self.sff_files]
rows = [sample.name for sample in self.samples]
data = [[sff.count_reads_in(sample) for sff in self.sff_files] for sample in self.samples]
self.frame = pandas.DataFrame(data, index=rows, columns=columns)
self.frame['Total'] = self.frame.sum(1)
# Regroup #
for sample in self.samples:
if len(sample.pieces) == 1: shutil.move(sample.pieces[0].path, sample.p.raw_sff)
else:
sh.sfffile("-o", sample.p.raw_sff, *[p.path for p in sample.pieces])
for p in sample.pieces: os.remove(p.path)
# Save report #
self.frame.to_csv(illumitag.view_dir + 'pyrosamples/report.csv')