def test_demultiplex_gz(tmp_dir, file_format):
"""
Test :py:func:`riboviz.demultiplex_fastq.demultiplex` using
GZIPped FASTQ files.
Each ``file_format`` consists of a FASTQ GZIP file name format and
the corresponding non-GZIP FASTQ file name format.
:param tmp_dir: Temporary directory
:type tmp_dir: str or unicode
:param file_format: File name format
:type file_format: tuple(str or unicode, str or unicode)
"""
gz_fmt, fmt = file_format
tmp_fastq_file = os.path.join(tmp_dir,
gz_fmt.format("test_multiplex"))
with open(os.path.join(riboviz.test.SIMDATA_DIR,
"multiplex.fastq"), "rb") as fr:
with gzip.open(tmp_fastq_file, "wb") as fw:
shutil.copyfileobj(fr, fw)
demultiplex_fastq.demultiplex(
os.path.join(riboviz.test.SIMDATA_DIR,
"multiplex_barcodes.tsv"),
tmp_fastq_file,
mismatches=2,
out_dir=tmp_dir)
actual_num_reads = os.path.join(
tmp_dir,
demultiplex_fastq.NUM_READS_FILE)
expected_num_reads = os.path.join(
riboviz.test.SIMDATA_DIR,
"deplex",
demultiplex_fastq.NUM_READS_FILE)
utils.equal_tsv(expected_num_reads, actual_num_reads)
for tag in ["Tag0", "Tag1", "Tag2", "Unassigned"]:
# Actual data has extension matching lower-case version
# of multiplexed file's extension.
actual_fq_gz = os.path.join(tmp_dir,
gz_fmt.lower().format(tag))
actual_fq = os.path.join(tmp_dir,
fmt.format(tag))
# Simulated data always has a .fastq extension.
expected_fq = os.path.join(riboviz.test.SIMDATA_DIR,
"deplex",
fastq.FASTQ_FORMAT.format(tag))
# Decompress actual_fq_gz
with gzip.open(actual_fq_gz, "rb") as fr:
with open(actual_fq, "wb") as fw:
shutil.copyfileobj(fr, fw)
fastq.equal_fastq(expected_fq, actual_fq)
# The definition of the simulated data means that Tag3 has no
# matches, as Tag0|1|2 will match any barcodes first. Check
# there is no Tag3-related output file.
assert not os.path.exists(os.path.join(tmp_dir,
gz_fmt.lower().format("Tag3")))
def test_demultiplex_output_error():
"""
Test :py:func:`riboviz.demultiplex_fastq.demultiplex` raises
``IOError`` if the output directory cannot be created.
"""
with pytest.raises(IOError):
demultiplex_fastq.demultiplex(
os.path.join(riboviz.test.SIMDATA_DIR, "multiplex_barcodes.tsv"),
os.path.join(riboviz.test.SIMDATA_DIR, "multiplex.fastq"),
# Pass existing file as out_dir value.
out_dir=os.path.join(riboviz.test.SIMDATA_DIR,
"multiplex_barcodes.tsv"))
def test_demultiplex_no_read1_file(tmp_dir):
"""
Test :py:func:`riboviz.demultiplex_fastq.demultiplex` raises
``FileNotFoundError`` if the FASTQ file is not found.
:param tmp_dir: Temporary directory
:type tmp_dir: str or unicode
"""
with pytest.raises(FileNotFoundError):
demultiplex_fastq.demultiplex(os.path.join(riboviz.test.SIMDATA_DIR,
"multiplex_barcodes.tsv"),
"nosuchfile.fastq",
out_dir=tmp_dir)
def test_demultiplex_no_sample_sheet(tmp_dir):
"""
Test :py:func:`riboviz.demultiplex_fastq.demultiplex` raises
``FileNotFoundError`` if the sample sheet is not found.
:param tmp_dir: Temporary directory
:type tmp_dir: str or unicode
"""
with pytest.raises(FileNotFoundError):
demultiplex_fastq.demultiplex("nosuchfile.tsv",
os.path.join(riboviz.test.SIMDATA_DIR,
"multiplex.fastq"),
out_dir=tmp_dir)
def test_demultiplex(tmp_dir, file_format):
"""
Test :py:func:`riboviz.demultiplex_fastq.demultiplex`.
:param tmp_dir: Temporary directory
:type tmp_dir: str or unicode
:param file_format: FASTQ file format
:type file_format: str or unicode
"""
tmp_fastq_file = os.path.join(tmp_dir,
file_format.format("test_multiplex"))
shutil.copyfile(os.path.join(riboviz.test.SIMDATA_DIR,
"multiplex.fastq"),
tmp_fastq_file)
demultiplex_fastq.demultiplex(
os.path.join(riboviz.test.SIMDATA_DIR,
"multiplex_barcodes.tsv"),
tmp_fastq_file,
mismatches=2,
out_dir=tmp_dir)
actual_num_reads = os.path.join(
tmp_dir,
demultiplex_fastq.NUM_READS_FILE)
expected_num_reads = os.path.join(
riboviz.test.SIMDATA_DIR,
"deplex",
demultiplex_fastq.NUM_READS_FILE)
utils.equal_tsv(expected_num_reads, actual_num_reads)
for tag in ["Tag0", "Tag1", "Tag2", "Unassigned"]:
# Actual data has extension matching lower-case version
# of multiplexed file's extension.
actual_fq = os.path.join(tmp_dir,
file_format.lower().format(tag))
# Simulated data always has a .fastq extension.
expected_fq = os.path.join(riboviz.test.SIMDATA_DIR,
"deplex",
fastq.FASTQ_FORMAT.format(tag))
fastq.equal_fastq(expected_fq, actual_fq)
# The definition of the simulated data means that Tag3 has no
# matches, as Tag0|1|2 will match any barcodes first. Check
# there is no Tag3-related output file.
assert not os.path.exists(os.path.join(tmp_dir,
file_format.lower().format("Tag3")))
def invoke_demultiplex_fastq():
"""
Parse command-line options then invoke
:py:func:`riboviz.demultiplex_fastq.demultiplex`.
"""
print(provenance.write_provenance_to_str(__file__))
options = parse_command_line_options()
sample_sheet_file = options.sample_sheet_file
read1_file = options.read1_file
read2_file = options.read2_file
mismatches = options.mismatches
out_dir = options.out_dir
delimiter = options.delimiter
demultiplex_fastq.demultiplex(sample_sheet_file,
read1_file,
read2_file,
mismatches,
out_dir,
delimiter)
