def test_check_seqs_no_primers(self):
"""check_seqs handles disabled primers """
# Fifth test, no primers, fixed length barcodes
# Should correct one of the passed barcodes
in_seqs = self.in_seqs_fixed_len_bc1
bc_map = self.bc_map_fixed_len_bc1
primer_seq_lens = {}
all_primers = {}
expected = self.expected_fasta_fixed_len_bc1_no_primers
out_f = FakeOutFile()
actual = check_seqs(
fasta_out=out_f,
fasta_files = [in_seqs],
starting_ix=0,
valid_map = bc_map,
qual_mappings={},
filters=[],
barcode_len=12,
keep_primer=False,
keep_barcode=False,
barcode_type="golay_12",
max_bc_errors=1.5,
remove_unassigned=True,
attempt_bc_correction=True,
primer_seqs_lens=primer_seq_lens,
all_primers=all_primers,
max_primer_mm=0,
disable_primer_check=True)
self.assertEqual(out_f.data,expected)
def test_check_seqs_fixed_len_bc(self):
"""check_seqs handles fixed length barcodes """
# Third test, fixed length barcodes, fixed length primers + one
# degenerate test. Should correct one of the passed barcodes
in_seqs = self.in_seqs_fixed_len_bc1
bc_map = self.bc_map_fixed_len_bc1
primer_seq_lens = self.primer_seq_lens_fixed_len_bc1
all_primers = self.all_primers_fixed_len_bc1
expected = self.expected_fasta_fixed_len_bc1
out_f = FakeOutFile()
actual = check_seqs(
fasta_out=out_f,
fasta_files = [in_seqs],
starting_ix=0,
valid_map = bc_map,
qual_mappings={},
filters=[],
barcode_len=12,
keep_primer=False,
keep_barcode=False,
barcode_type="golay_12",
max_bc_errors=1.5,
remove_unassigned=True,
attempt_bc_correction=True,
primer_seqs_lens=primer_seq_lens,
all_primers=all_primers,
max_primer_mm=0,
disable_primer_check=False)
self.assertEqual(out_f.data,expected)
# Fourth test-set max_bc_errors to 0, and allow some primer mismatches
in_seqs = self.in_seqs_fixed_len_bc2
bc_map = self.bc_map_fixed_len_bc2
primer_seq_lens = self.primer_seq_lens_fixed_len_bc2
all_primers = self.all_primers_fixed_len_bc2
expected = self.expected_fasta_fixed_len_bc2
out_f = FakeOutFile()
actual = check_seqs(
fasta_out=out_f,
fasta_files = [in_seqs],
starting_ix=0,
valid_map = bc_map,
qual_mappings={},
filters=[],
barcode_len=12,
keep_primer=False,
keep_barcode=False,
barcode_type="golay_12",
max_bc_errors=0.5,
remove_unassigned=True,
attempt_bc_correction=True,
primer_seqs_lens=primer_seq_lens,
all_primers=all_primers,
max_primer_mm=1,
disable_primer_check=False)
self.assertEqual(out_f.data,expected)
def test_check_seqs_variable_len_bc(self):
"""check_seqs handles variable length barcodes """
# Simple test with variable length primers
in_seqs = self.in_seqs_variable_len_bc1
bc_map = self.bc_map_variable_len_bc1
primer_seq_lens = self.primer_seq_lens_variable_len_bc1
all_primers = self.all_primers_variable_len_bc1
expected = self.expected_fasta_variable_len_bc1
out_f = FakeOutFile()
actual = check_seqs(
fasta_out=out_f,
fasta_files = [in_seqs],
starting_ix=0,
valid_map = bc_map,
qual_mappings={},
filters=[],
barcode_len=None,
keep_primer=False,
keep_barcode=False,
barcode_type="variable_length",
max_bc_errors=0,
remove_unassigned=True,
attempt_bc_correction=False,
primer_seqs_lens=primer_seq_lens,
all_primers=all_primers,
max_primer_mm=0,
disable_primer_check=False)
self.assertEqual(out_f.data,expected)
# Second test, includes truncated form of one of the barcodes-the
# longest barcode should be found first
in_seqs = self.in_seqs_variable_len_bc2
bc_map = self.bc_map_variable_len_bc2
primer_seq_lens = self.primer_seq_lens_variable_len_bc2
all_primers = self.all_primers_variable_len_bc2
expected = self.expected_fasta_variable_len_bc2
out_f = FakeOutFile()
actual = check_seqs(
fasta_out=out_f,
fasta_files = [in_seqs],
starting_ix=0,
valid_map = bc_map,
qual_mappings={},
filters=[],
barcode_len=None,
keep_primer=False,
keep_barcode=False,
barcode_type="variable_length",
max_bc_errors=0,
remove_unassigned=True,
attempt_bc_correction=False,
primer_seqs_lens=primer_seq_lens,
all_primers=all_primers,
max_primer_mm=0,
disable_primer_check=False)
self.assertEqual(out_f.data,expected)
