def test_process_barcode_paired_end_data(self):
""" Handles paired fastq lines, parses barcodes """
fastq1_data = ["HWI-ST830", "AAAATTTTCCCCGGGG", "1234567890ABCDEF"]
fastq2_data = ["HWI-ST830", "TCCCCGGGG", "ABCDEFGHI"]
reads1_out = FakeOutFile()
reads2_out = FakeOutFile()
bcs_out = FakeOutFile()
process_barcode_paired_end_data(fastq1_data, fastq2_data,
bcs_out, reads1_out, reads2_out, bc1_len=5, bc2_len=3,
rev_comp_bc1=True, rev_comp_bc2=True)
actual_bcs = bcs_out.data.split('\n')
expected_bcs = ['@HWI-ST830', 'ATTTTGGA', '+', '54321CBA', '']
self.assertEqual(actual_bcs, expected_bcs)
actual_reads = reads1_out.data.split('\n')
expected_reads = ['@HWI-ST830', 'TTTCCCCGGGG', '+', '67890ABCDEF', '']
self.assertEqual(actual_reads, expected_reads)
actual_reads = reads2_out.data.split('\n')
expected_reads = ['@HWI-ST830', 'CCGGGG', '+', 'DEFGHI', '']
self.assertEqual(actual_reads, expected_reads)
def test_process_barcode_paired_end_data(self):
""" Handles paired fastq lines, parses barcodes """
fastq1_data = ["HWI-ST830", "AAAATTTTCCCCGGGG",
np.arange(3, 19, dtype=np.int8)]
fastq2_data = ["HWI-ST830", "TCCCCGGGG", np.arange(3, 12, dtype=np.int8)]
reads1_out = FakeOutFile()
reads2_out = FakeOutFile()
bcs_out = FakeOutFile()
process_barcode_paired_end_data(fastq1_data, fastq2_data,
bcs_out, reads1_out, reads2_out, bc1_len=5, bc2_len=3,
rev_comp_bc1=True, rev_comp_bc2=True)
actual_bcs = bcs_out.data.split('\n')
expected_bcs = ['@HWI-ST830', 'ATTTTGGA', '+', "('&%$&%$", '']
self.assertEqual(actual_bcs, expected_bcs)
actual_reads = reads1_out.data.split('\n')
expected_reads = ['@HWI-ST830', 'TTTCCCCGGGG', '+', ')*+,-./0123', '']
self.assertEqual(actual_reads, expected_reads)
actual_reads = reads2_out.data.split('\n')
expected_reads = ['@HWI-ST830', 'CCGGGG', '+', "'()*+,", '']
self.assertEqual(actual_reads, expected_reads)
def test_process_barcode_paired_end_data(self):
""" Handles paired fastq lines, parses barcodes """
fastq1_data = ["HWI-ST830", "AAAATTTTCCCCGGGG", "1234567890ABCDEF"]
fastq2_data = ["HWI-ST830", "TCCCCGGGG", "ABCDEFGHI"]
reads1_out = FakeOutFile()
reads2_out = FakeOutFile()
bcs_out = FakeOutFile()
process_barcode_paired_end_data(fastq1_data,
fastq2_data,
bcs_out,
reads1_out,
reads2_out,
bc1_len=5,
bc2_len=3,
rev_comp_bc1=True,
rev_comp_bc2=True)
actual_bcs = bcs_out.data.split('\n')
expected_bcs = ['@HWI-ST830', 'ATTTTGGA', '+', '54321CBA', '']
self.assertEqual(actual_bcs, expected_bcs)
actual_reads = reads1_out.data.split('\n')
expected_reads = ['@HWI-ST830', 'TTTCCCCGGGG', '+', '67890ABCDEF', '']
self.assertEqual(actual_reads, expected_reads)
actual_reads = reads2_out.data.split('\n')
expected_reads = ['@HWI-ST830', 'CCGGGG', '+', 'DEFGHI', '']
self.assertEqual(actual_reads, expected_reads)
def test_process_barcode_paired_end_data_orientation_reverse_in_read1(
self):
""" Handles paired fastq lines, parses barcodes, orients reads """
fastq1_data = ["HWI-ST830", "ATCGATCGATCGATCGATCG",
"1234567890ABCDEFGHIJ"]
fastq2_data = ["HWI-ST830", "GGTTCCAA", "ABCDEFGH"]
reads1_out = FakeOutFile()
reads2_out = FakeOutFile()
bcs_out = FakeOutFile()
forward_primers = [compile(''.join([self.iupac[symbol] for
symbol in 'TTTTT']))]
reverse_primers = [compile(''.join([self.iupac[symbol] for
symbol in 'CGATCGA']))]
output_bc_not_oriented = FakeOutFile()
fastq1_out_not_oriented = FakeOutFile()
fastq2_out_not_oriented = FakeOutFile()
# With a forward primer match in read 2, should reverse read order
process_barcode_paired_end_data(fastq1_data, fastq2_data,
bcs_out, reads1_out, reads2_out, bc1_len=5, bc2_len=3,
rev_comp_bc1=False, rev_comp_bc2=False,
attempt_read_orientation=True, forward_primers=forward_primers,
reverse_primers=reverse_primers,
output_bc_not_oriented=output_bc_not_oriented,
fastq1_out_not_oriented=fastq1_out_not_oriented,
fastq2_out_not_oriented=fastq2_out_not_oriented)
actual_bcs = bcs_out.data.split('\n')
expected_bcs = ['@HWI-ST830', 'GGTTCATC', '+', 'ABCDE123', '']
self.assertEqual(actual_bcs, expected_bcs)
actual_reads = reads1_out.data.split('\n')
expected_reads = ['@HWI-ST830', 'CAA', '+', 'FGH', '']
self.assertEqual(actual_reads, expected_reads)
actual_reads = reads2_out.data.split('\n')
expected_reads = ['@HWI-ST830', 'GATCGATCGATCGATCG', '+',
'4567890ABCDEFGHIJ', '']
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)
actual_reads_not_oriented = fastq2_out_not_oriented.data.split('\n')
expected_reads = ['']
self.assertEqual(actual_reads_not_oriented, expected_reads)
def test_process_barcode_paired_end_data_orientation_no_match(self):
""" Handles paired fastq lines, parses barcodes, orients reads """
fastq1_data = ["HWI-ST830", "ATCGATCGATCGATCGATCG",
"1234567890ABCDEFGHIJ"]
fastq2_data = ["HWI-ST830", "GGTTCCAA", "ABCDEFGH"]
reads1_out = FakeOutFile()
reads2_out = FakeOutFile()
bcs_out = FakeOutFile()
forward_primers = [compile(''.join([self.iupac[symbol] for
symbol in 'AYA']))]
reverse_primers = [compile(''.join([self.iupac[symbol] for
symbol in 'ATA']))]
output_bc_not_oriented = FakeOutFile()
fastq1_out_not_oriented = FakeOutFile()
fastq2_out_not_oriented = FakeOutFile()
# With no matches, should write to the not_oriented files, and keep
# in the same order of file 1 and file 2
process_barcode_paired_end_data(fastq1_data, fastq2_data,
bcs_out, reads1_out, reads2_out, bc1_len=5, bc2_len=3,
rev_comp_bc1=False, rev_comp_bc2=False,
attempt_read_orientation=True, forward_primers=forward_primers,
reverse_primers=reverse_primers,
output_bc_not_oriented=output_bc_not_oriented,
fastq1_out_not_oriented=fastq1_out_not_oriented,
fastq2_out_not_oriented=fastq2_out_not_oriented)
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_reads = reads2_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', 'ATCGAGGT', '+', '12345ABC', '']
self.assertEqual(actual_bcs_not_oriented, expected_bcs)
actual_reads_not_oriented = fastq1_out_not_oriented.data.split('\n')
expected_reads = ['@HWI-ST830', 'TCGATCGATCGATCG', '+',
'67890ABCDEFGHIJ', '']
self.assertEqual(actual_reads_not_oriented, expected_reads)
actual_reads_not_oriented = fastq2_out_not_oriented.data.split('\n')
expected_reads = ['@HWI-ST830', 'TCCAA', '+', 'DEFGH', '']
self.assertEqual(actual_reads_not_oriented, expected_reads)
def test_process_barcode_paired_end_data_orientation_reverse_in_read1(
self):
""" Handles paired fastq lines, parses barcodes, orients reads """
fastq1_data = ["HWI-ST830", "ATCGATCGATCGATCGATCG",
np.arange(3, 23, dtype=np.int8)]
fastq2_data = ["HWI-ST830", "GGTTCCAA", np.arange(3, 11, dtype=np.int8)]
reads1_out = FakeOutFile()
reads2_out = FakeOutFile()
bcs_out = FakeOutFile()
forward_primers = [compile(''.join([self.iupac[symbol] for
symbol in 'TTTTT']))]
reverse_primers = [compile(''.join([self.iupac[symbol] for
symbol in 'CGATCGA']))]
output_bc_not_oriented = FakeOutFile()
fastq1_out_not_oriented = FakeOutFile()
fastq2_out_not_oriented = FakeOutFile()
# With a forward primer match in read 2, should reverse read order
process_barcode_paired_end_data(fastq1_data, fastq2_data,
bcs_out, reads1_out, reads2_out, bc1_len=5, bc2_len=3,
rev_comp_bc1=False, rev_comp_bc2=False,
attempt_read_orientation=True, forward_primers=forward_primers,
reverse_primers=reverse_primers,
output_bc_not_oriented=output_bc_not_oriented,
fastq1_out_not_oriented=fastq1_out_not_oriented,
fastq2_out_not_oriented=fastq2_out_not_oriented)
actual_bcs = bcs_out.data.split('\n')
expected_bcs = ['@HWI-ST830', 'GGTTCATC', '+', "$%&'($%&", '']
self.assertEqual(actual_bcs, expected_bcs)
actual_reads = reads1_out.data.split('\n')
expected_reads = ['@HWI-ST830', 'CAA', '+', ')*+', '']
self.assertEqual(actual_reads, expected_reads)
actual_reads = reads2_out.data.split('\n')
expected_reads = ['@HWI-ST830', 'GATCGATCGATCGATCG', '+',
"'()*+,-./01234567", '']
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)
actual_reads_not_oriented = fastq2_out_not_oriented.data.split('\n')
expected_reads = ['']
self.assertEqual(actual_reads_not_oriented, expected_reads)
def test_process_barcode_paired_end_data_orientation_forward_in_read2(
self):
""" Handles paired fastq lines, parses barcodes, orients reads """
fastq1_data = ["HWI-ST830", "ATCGATCGATCGATCGATCG",
np.arange(3, 23, dtype=np.int8)]
fastq2_data = ["HWI-ST830", "GGTTCCAA", np.arange(3, 11, dtype=np.int8)]
reads1_out = FakeOutFile()
reads2_out = FakeOutFile()
bcs_out = FakeOutFile()
forward_primers = [compile(''.join([self.iupac[symbol] for
symbol in 'TTCCA']))]
reverse_primers = [compile(''.join([self.iupac[symbol] for
symbol in 'ATA']))]
output_bc_not_oriented = FakeOutFile()
fastq1_out_not_oriented = FakeOutFile()
fastq2_out_not_oriented = FakeOutFile()
# With a forward primer match in read 2, should reverse read order
process_barcode_paired_end_data(fastq1_data, fastq2_data,
bcs_out, reads1_out, reads2_out, bc1_len=5, bc2_len=3,
rev_comp_bc1=False, rev_comp_bc2=False,
attempt_read_orientation=True, forward_primers=forward_primers,
reverse_primers=reverse_primers,
output_bc_not_oriented=output_bc_not_oriented,
fastq1_out_not_oriented=fastq1_out_not_oriented,
fastq2_out_not_oriented=fastq2_out_not_oriented)
actual_bcs = bcs_out.data.split('\n')
expected_bcs = ['@HWI-ST830', 'GGTTCATC', '+', "$%&'($%&", '']
self.assertEqual(actual_bcs, expected_bcs)
actual_reads = reads1_out.data.split('\n')
expected_reads = ['@HWI-ST830', 'CAA', '+', ')*+', '']
self.assertEqual(actual_reads, expected_reads)
actual_reads = reads2_out.data.split('\n')
expected_reads = ['@HWI-ST830', 'GATCGATCGATCGATCG', '+',
"'()*+,-./01234567", '']
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)
actual_reads_not_oriented = fastq2_out_not_oriented.data.split('\n')
expected_reads = ['']
self.assertEqual(actual_reads_not_oriented, expected_reads)
def test_process_barcode_paired_end_data_orientation_rev_in_read2(self):
""" Handles paired fastq lines, parses barcodes, orients reads """
fastq1_data = [
"HWI-ST830", "ATCGATCGATCGATCGATCG", "1234567890ABCDEFGHIJ"
]
fastq2_data = ["HWI-ST830", "GGTTCCAA", "ABCDEFGH"]
reads1_out = FakeOutFile()
reads2_out = FakeOutFile()
bcs_out = FakeOutFile()
forward_primers = [
compile(''.join([self.iupac[symbol] for symbol in 'TTTTTT']))
]
reverse_primers = [
compile(''.join([self.iupac[symbol] for symbol in 'TCCAA']))
]
output_bc_not_oriented = FakeOutFile()
fastq1_out_not_oriented = FakeOutFile()
fastq2_out_not_oriented = FakeOutFile()
# With a reverse primer in read 2, should write in current order.
process_barcode_paired_end_data(
fastq1_data,
fastq2_data,
bcs_out,
reads1_out,
reads2_out,
bc1_len=5,
bc2_len=3,
rev_comp_bc1=False,
rev_comp_bc2=False,
attempt_read_orientation=True,
forward_primers=forward_primers,
reverse_primers=reverse_primers,
output_bc_not_oriented=output_bc_not_oriented,
fastq1_out_not_oriented=fastq1_out_not_oriented,
fastq2_out_not_oriented=fastq2_out_not_oriented)
actual_bcs = bcs_out.data.split('\n')
expected_bcs = ['@HWI-ST830', 'ATCGAGGT', '+', '12345ABC', '']
self.assertEqual(actual_bcs, expected_bcs)
actual_reads = reads1_out.data.split('\n')
expected_reads = [
'@HWI-ST830', 'TCGATCGATCGATCG', '+', '67890ABCDEFGHIJ', ''
]
self.assertEqual(actual_reads, expected_reads)
actual_reads = reads2_out.data.split('\n')
expected_reads = ['@HWI-ST830', 'TCCAA', '+', 'DEFGH', '']
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)
actual_reads_not_oriented = fastq2_out_not_oriented.data.split('\n')
expected_reads = ['']
self.assertEqual(actual_reads_not_oriented, expected_reads)
