def test_extract_barcodes_paired_end(self):
""" Extracts barcodes from paired-end reads """
fastq1_lines =\
"@HWI-ST830\nAAAATTTTCCCCGGGG\n+\n1234567890ABCDEF".split('\n')
fastq2_lines =\
"@HWI-ST830\nGGGGTTTTAAAACCCC\n+\n1234567890ABCDEF".split('\n')
extract_barcodes(fastq1 = fastq1_lines, fastq2 = fastq2_lines,
input_type = "barcode_paired_end", output_dir = self.output_dir)
output_bcs_fp = open(join(self.output_dir, "barcodes.fastq"), "U")
actual_bcs = [line for line in output_bcs_fp]
expected_bcs =\
['@HWI-ST830\n', 'AAAATTGGGGTT\n', '+\n', '123456123456\n']
self.assertEqual(actual_bcs, expected_bcs)
# reads 1 output
output_reads_fp = open(join(self.output_dir, "reads1.fastq"), "U")
actual_reads = [line for line in output_reads_fp]
expected_reads = ['@HWI-ST830\n', 'TTCCCCGGGG\n', '+\n', '7890ABCDEF\n']
self.assertEqual(actual_reads, expected_reads)
# reads 2 output
output_reads_fp = open(join(self.output_dir, "reads2.fastq"), "U")
actual_reads = [line for line in output_reads_fp]
expected_reads = ['@HWI-ST830\n', 'TTAAAACCCC\n', '+\n', '7890ABCDEF\n']
self.assertEqual(actual_reads, expected_reads)
def main():
option_parser, opts, args =\
parse_command_line_parameters(**script_info)
if opts.attempt_read_reorientation:
if not opts.mapping_fp:
option_parser.error("To use --attempt_read_reorientation, one must "
"supply a mapping file that contains both LinkerPrimerSequence "
"and ReversePrimer columns.")
if opts.input_type == "barcode_paired_end":
if not opts.fastq2:
option_parser.error("To use input_type of barcode_paired_end, "
"a second fastq file must be specified with --fastq2")
if not opts.fastq2:
disable_header_match = True
else:
disable_header_match = opts.disable_header_match
fastq1 = qiime_open(opts.fastq1)
if opts.fastq2:
fastq2 = qiime_open(opts.fastq2)
else:
fastq2 = None
create_dir(opts.output_dir)
if opts.mapping_fp:
map_fp = qiime_open(opts.mapping_fp)
else:
map_fp = None
extract_barcodes(fastq1, fastq2, opts.output_dir, opts.input_type,
opts.bc1_len, opts.bc2_len, opts.rev_comp_bc1, opts.rev_comp_bc2,
opts.char_delineator, opts.switch_bc_order, map_fp,
opts.attempt_read_reorientation, disable_header_match)
def main():
option_parser, opts, args =\
parse_command_line_parameters(**script_info)
if opts.attempt_read_reorientation:
if not opts.mapping_fp:
option_parser.error("To use --attempt_read_reorientation, one must "
"supply a mapping file that contains both LinkerPrimerSequence "
"and ReversePrimer columns.")
if opts.input_type == "barcode_paired_end":
if not opts.fastq2:
option_parser.error("To use input_type of barcode_paired_end, "
"a second fastq file must be specified with --fastq2")
if not opts.fastq2:
disable_header_match = True
else:
disable_header_match = opts.disable_header_match
fastq1 = qiime_open(opts.fastq1)
if opts.fastq2:
fastq2 = qiime_open(opts.fastq2)
else:
fastq2 = None
create_dir(opts.output_dir)
if opts.mapping_fp:
map_fp = qiime_open(opts.mapping_fp)
else:
map_fp = None
extract_barcodes(fastq1, fastq2, opts.output_dir, opts.input_type,
opts.bc1_len, opts.bc2_len, opts.rev_comp_bc1, opts.rev_comp_bc2,
opts.char_delineator, opts.switch_bc_order, map_fp,
opts.attempt_read_reorientation, disable_header_match)
def test_extract_barcodes_from_labels(self):
""" Extracts barcodes from fastq labels """
fastq_lines =\
"@HWI-ST830:GTATCT\nAAAATTTTCCCCGGGG\n+\n1234567890ABCDEF".split('\n')
extract_barcodes(fastq_lines, input_type="barcode_in_label",
output_dir=self.output_dir, disable_header_match=True)
output_bcs_fp = open(join(self.output_dir, "barcodes.fastq"), "U")
actual_bcs = [line for line in output_bcs_fp]
expected_bcs =\
['@HWI-ST830:GTATCT\n', 'GTATCT\n', '+\n', 'FFFFFF\n']
self.assertEqual(actual_bcs, expected_bcs)
def test_extract_barcodes_from_labels(self):
""" Extracts barcodes from fastq labels """
fastq_lines =\
"@HWI-ST830:GTATCT\nAAAATTTTCCCCGGGG\n+\n1234567890ABCDEF".split('\n')
extract_barcodes(fastq_lines, input_type="barcode_in_label",
output_dir=self.output_dir, disable_header_match=True)
output_bcs_fp = open(join(self.output_dir, "barcodes.fastq"), "U")
actual_bcs = [line for line in output_bcs_fp]
expected_bcs =\
['@HWI-ST830:GTATCT\n', 'GTATCT\n', '+\n', "''''''\n"]
self.assertEqual(actual_bcs, expected_bcs)
def test_extract_barcodes_single_end(self):
""" Extracts barcodes from single-end read """
fastq_lines =\
"@HWI-ST830\nAAAATTTTCCCCGGGG\n+\n1234567890ABCDEF".split('\n')
extract_barcodes(fastq_lines, output_dir = self.output_dir,
disable_header_match = True)
output_bcs_fp = open(join(self.output_dir, "barcodes.fastq"), "U")
actual_bcs = [line for line in output_bcs_fp]
expected_bcs = ['@HWI-ST830\n', 'AAAATT\n', '+\n', '123456\n']
self.assertEqual(actual_bcs, expected_bcs)
output_reads_fp = open(join(self.output_dir, "reads.fastq"), "U")
actual_reads = [line for line in output_reads_fp]
expected_reads = ['@HWI-ST830\n', 'TTCCCCGGGG\n', '+\n', '7890ABCDEF\n']
self.assertEqual(actual_reads, expected_reads)
def test_extract_barcodes_stitched_reads(self):
""" Extracts barcodes from ends of a single read """
fastq_lines =\
"@HWI-ST830\nAAAATTTTCCCCGGGG\n+\n1234567890ABCDEF\n".split('\n')
extract_barcodes(fastq_lines, input_type="barcode_paired_stitched",
output_dir=self.output_dir, disable_header_match=True)
output_bcs_fp = open(join(self.output_dir, "barcodes.fastq"), "U")
actual_bcs = [line for line in output_bcs_fp]
expected_bcs =\
['@HWI-ST830\n', 'AAAATTCCGGGG\n', '+\n', '123456ABCDEF\n']
self.assertEqual(actual_bcs, expected_bcs)
output_reads_fp = open(join(self.output_dir, "reads.fastq"), "U")
actual_reads = [line for line in output_reads_fp]
expected_reads = ['@HWI-ST830\n', 'TTCC\n', '+\n', '7890\n']
self.assertEqual(actual_reads, expected_reads)
def test_extract_barcodes_paired_end(self):
""" Extracts barcodes from paired-end reads """
fastq1_lines =\
"@HWI-ST830\nAAAATTTTCCCCGGGG\n+\n1234567890ABCDEF".split('\n')
fastq2_lines =\
"@HWI-ST830\nGGGGTTTTAAAACCCC\n+\n1234567890ABCDEF".split('\n')
extract_barcodes(fastq1=fastq1_lines, fastq2=fastq2_lines,
input_type="barcode_paired_end", output_dir=self.output_dir)
output_bcs_fp = open(join(self.output_dir, "barcodes.fastq"), "U")
actual_bcs = [line for line in output_bcs_fp]
expected_bcs =\
['@HWI-ST830\n', 'AAAATTGGGGTT\n', '+\n', '123456123456\n']
self.assertEqual(actual_bcs, expected_bcs)
# reads 1 output
output_reads_fp = open(join(self.output_dir, "reads1.fastq"), "U")
actual_reads = [line for line in output_reads_fp]
expected_reads = [
'@HWI-ST830\n',
'TTCCCCGGGG\n',
'+\n',
'7890ABCDEF\n']
self.assertEqual(actual_reads, expected_reads)
# reads 2 output
output_reads_fp = open(join(self.output_dir, "reads2.fastq"), "U")
actual_reads = [line for line in output_reads_fp]
expected_reads = [
'@HWI-ST830\n',
'TTAAAACCCC\n',
'+\n',
'7890ABCDEF\n']
self.assertEqual(actual_reads, expected_reads)