def test_process_barcode_paired_stitched_reverse_primer_match(self):
""" Handles stitched barcode data, parses barcodes from ends """
fastq1_data = [
"HWI-ST830", "ATCGATCGATCGATCGATCG",
np.arange(3, 23, dtype=np.int8)
]
reads1_out = FakeOutFile()
bcs_out = FakeOutFile()
forward_primers = [
compile(''.join([self.iupac[symbol] for symbol in 'AAAAAA']))
]
reverse_primers = [
compile(''.join([self.iupac[symbol] for symbol in 'GATCG']))
]
output_bc_not_oriented = FakeOutFile()
fastq1_out_not_oriented = FakeOutFile()
# With reverse primer match, should write in order of read2, read 1
process_barcode_paired_stitched(
fastq1_data,
bcs_out,
reads1_out,
bc1_len=3,
bc2_len=4,
rev_comp_bc1=True,
rev_comp_bc2=False,
attempt_read_orientation=True,
forward_primers=forward_primers,
reverse_primers=reverse_primers,
output_bc_not_oriented=output_bc_not_oriented,
fastq_out_not_oriented=fastq1_out_not_oriented,
switch_bc_order=False)
actual_bcs = bcs_out.data.split('\n')
expected_bcs = ['@HWI-ST830', 'TCGCGAT', '+', "567'&%$", '']
self.assertEqual(actual_bcs, expected_bcs)
actual_reads = reads1_out.data.split('\n')
expected_reads = [
'@HWI-ST830', 'TCGATCGATCGAT', '+', '43210/.-,+*)(', ''
]
self.assertEqual(actual_reads, expected_reads)
actual_bcs_not_oriented = output_bc_not_oriented.data.split('\n')
expected_bcs = ['']
self.assertEqual(actual_bcs_not_oriented, expected_bcs)
actual_reads_not_oriented = fastq1_out_not_oriented.data.split('\n')
expected_reads = ['']
self.assertEqual(actual_reads_not_oriented, expected_reads)
def test_process_barcode_paired_stitched(self):
""" Handles stitched barcode data, parses barcodes from ends """
fastq1_data = [
"HWI-ST830", "ATCGATCGATCGATCGATCG",
np.arange(3, 23, dtype=np.int8)
]
reads1_out = FakeOutFile()
bcs_out = FakeOutFile()
forward_primers = [
compile(''.join([self.iupac[symbol] for symbol in 'ATA']))
]
reverse_primers = [
compile(''.join([self.iupac[symbol] for symbol in 'ATA']))
]
output_bc_not_oriented = FakeOutFile()
fastq1_out_not_oriented = FakeOutFile()
# With no matches, should write to the not_oriented files, and keep
# in the same order of output file
process_barcode_paired_stitched(
fastq1_data,
bcs_out,
reads1_out,
bc1_len=3,
bc2_len=4,
rev_comp_bc1=True,
rev_comp_bc2=True,
attempt_read_orientation=True,
forward_primers=forward_primers,
reverse_primers=reverse_primers,
output_bc_not_oriented=output_bc_not_oriented,
fastq_out_not_oriented=fastq1_out_not_oriented,
switch_bc_order=True)
actual_bcs = bcs_out.data.split('\n')
expected_bcs = ['']
self.assertEqual(actual_bcs, expected_bcs)
actual_reads = reads1_out.data.split('\n')
expected_reads = ['']
self.assertEqual(actual_reads, expected_reads)
actual_bcs_not_oriented = output_bc_not_oriented.data.split('\n')
expected_bcs = ['@HWI-ST830', 'CGATGAT', '+', '7654&%$', '']
self.assertEqual(actual_bcs_not_oriented, expected_bcs)
actual_reads_not_oriented = fastq1_out_not_oriented.data.split('\n')
expected_reads =\
['@HWI-ST830', 'GATCGATCGATCG', '+', "'()*+,-./0123", '']
self.assertEqual(actual_reads_not_oriented, expected_reads)
def test_process_barcode_paired_stitched_forward_primer_match(self):
""" Handles stitched barcode data, parses barcodes from ends """
fastq1_data = [
"HWI-ST830", "ATCGATCGATCGATCGATCG", "1234567890ABCDEFGHIJ"
]
reads1_out = FakeOutFile()
bcs_out = FakeOutFile()
forward_primers = [
compile(''.join([self.iupac[symbol] for symbol in 'GATCGA']))
]
reverse_primers = [
compile(''.join([self.iupac[symbol] for symbol in 'ATA']))
]
output_bc_not_oriented = FakeOutFile()
fastq1_out_not_oriented = FakeOutFile()
# With forward primer match, should write in order of read 1, read 2
process_barcode_paired_stitched(
fastq1_data,
bcs_out,
reads1_out,
bc1_len=3,
bc2_len=4,
rev_comp_bc1=True,
rev_comp_bc2=True,
attempt_read_orientation=True,
forward_primers=forward_primers,
reverse_primers=reverse_primers,
output_bc_not_oriented=output_bc_not_oriented,
fastq_out_not_oriented=fastq1_out_not_oriented,
switch_bc_order=True)
actual_bcs = bcs_out.data.split('\n')
expected_bcs = ['@HWI-ST830', 'CGATGAT', '+', 'JIHG321', '']
self.assertEqual(actual_bcs, expected_bcs)
actual_reads = reads1_out.data.split('\n')
expected_reads = [
'@HWI-ST830', 'GATCGATCGATCG', '+', '4567890ABCDEF', ''
]
self.assertEqual(actual_reads, expected_reads)
actual_bcs_not_oriented = output_bc_not_oriented.data.split('\n')
expected_bcs = ['']
self.assertEqual(actual_bcs_not_oriented, expected_bcs)
actual_reads_not_oriented = fastq1_out_not_oriented.data.split('\n')
expected_reads = ['']
self.assertEqual(actual_reads_not_oriented, expected_reads)
def test_process_barcode_paired_stitched(self):
""" Handles stitched barcode data, parses barcodes from ends """
fastq1_data = ["HWI-ST830","ATCGATCGATCGATCGATCG",
"1234567890ABCDEFGHIJ"]
reads1_out = FakeOutFile()
bcs_out = FakeOutFile()
forward_primers = [compile(''.join([self.iupac[symbol] for\
symbol in 'ATA']))]
reverse_primers = [compile(''.join([self.iupac[symbol] for\
symbol in 'ATA']))]
output_bc_not_oriented = FakeOutFile()
fastq1_out_not_oriented = FakeOutFile()
# With no matches, should write to the not_oriented files, and keep
# in the same order of output file
process_barcode_paired_stitched(fastq1_data,
bcs_out, reads1_out, bc1_len=3, bc2_len=4,
rev_comp_bc1=True, rev_comp_bc2=True,
attempt_read_orientation = True,
forward_primers = forward_primers,
reverse_primers = reverse_primers,
output_bc_not_oriented = output_bc_not_oriented,
fastq_out_not_oriented = fastq1_out_not_oriented,
switch_bc_order = True)
actual_bcs = bcs_out.data.split('\n')
expected_bcs = ['']
self.assertEqual(actual_bcs, expected_bcs)
actual_reads = reads1_out.data.split('\n')
expected_reads = ['']
self.assertEqual(actual_reads, expected_reads)
actual_bcs_not_oriented = output_bc_not_oriented.data.split('\n')
expected_bcs = ['@HWI-ST830', 'CGATGAT', '+', 'JIHG321', '']
self.assertEqual(actual_bcs_not_oriented, expected_bcs)
actual_reads_not_oriented = fastq1_out_not_oriented.data.split('\n')
expected_reads =\
['@HWI-ST830', 'GATCGATCGATCG', '+', '4567890ABCDEF', '']
self.assertEqual(actual_reads_not_oriented, expected_reads)
def test_process_barcode_paired_stitched_reverse_primer_match(self):
""" Handles stitched barcode data, parses barcodes from ends """
fastq1_data = ["HWI-ST830","ATCGATCGATCGATCGATCG",
"1234567890ABCDEFGHIJ"]
reads1_out = FakeOutFile()
bcs_out = FakeOutFile()
forward_primers = [compile(''.join([self.iupac[symbol] for\
symbol in 'AAAAAA']))]
reverse_primers = [compile(''.join([self.iupac[symbol] for\
symbol in 'GATCG']))]
output_bc_not_oriented = FakeOutFile()
fastq1_out_not_oriented = FakeOutFile()
# With reverse primer match, should write in order of read2, read 1
process_barcode_paired_stitched(fastq1_data,
bcs_out, reads1_out, bc1_len=3, bc2_len=4,
rev_comp_bc1=True, rev_comp_bc2=False,
attempt_read_orientation = True,
forward_primers = forward_primers,
reverse_primers = reverse_primers,
output_bc_not_oriented = output_bc_not_oriented,
fastq_out_not_oriented = fastq1_out_not_oriented,
switch_bc_order = False)
actual_bcs = bcs_out.data.split('\n')
expected_bcs = ['@HWI-ST830', 'TCGCGAT', '+', 'HIJ4321', '']
self.assertEqual(actual_bcs, expected_bcs)
actual_reads = reads1_out.data.split('\n')
expected_reads = ['@HWI-ST830', 'TCGATCGATCGAT', '+',
'GFEDCBA098765', '']
self.assertEqual(actual_reads, expected_reads)
actual_bcs_not_oriented = output_bc_not_oriented.data.split('\n')
expected_bcs = ['']
self.assertEqual(actual_bcs_not_oriented, expected_bcs)
actual_reads_not_oriented = fastq1_out_not_oriented.data.split('\n')
expected_reads = ['']
self.assertEqual(actual_reads_not_oriented, expected_reads)
def test_process_barcode_paired_stitched_forward_primer_match(self):
""" Handles stitched barcode data, parses barcodes from ends """
fastq1_data = ["HWI-ST830", "ATCGATCGATCGATCGATCG",
np.arange(3, 23, dtype=np.int8)]
reads1_out = FakeOutFile()
bcs_out = FakeOutFile()
forward_primers = [compile(''.join([self.iupac[symbol] for
symbol in 'GATCGA']))]
reverse_primers = [compile(''.join([self.iupac[symbol] for
symbol in 'ATA']))]
output_bc_not_oriented = FakeOutFile()
fastq1_out_not_oriented = FakeOutFile()
# With forward primer match, should write in order of read 1, read 2
process_barcode_paired_stitched(fastq1_data,
bcs_out, reads1_out, bc1_len=3, bc2_len=4,
rev_comp_bc1=True, rev_comp_bc2=True,
attempt_read_orientation=True,
forward_primers=forward_primers,
reverse_primers=reverse_primers,
output_bc_not_oriented=output_bc_not_oriented,
fastq_out_not_oriented=fastq1_out_not_oriented,
switch_bc_order=True)
actual_bcs = bcs_out.data.split('\n')
expected_bcs = ['@HWI-ST830', 'CGATGAT', '+', '7654&%$', '']
self.assertEqual(actual_bcs, expected_bcs)
actual_reads = reads1_out.data.split('\n')
expected_reads = ['@HWI-ST830', 'GATCGATCGATCG', '+',
"'()*+,-./0123", '']
self.assertEqual(actual_reads, expected_reads)
actual_bcs_not_oriented = output_bc_not_oriented.data.split('\n')
expected_bcs = ['']
self.assertEqual(actual_bcs_not_oriented, expected_bcs)
actual_reads_not_oriented = fastq1_out_not_oriented.data.split('\n')
expected_reads = ['']
self.assertEqual(actual_reads_not_oriented, expected_reads)
