def test_prepare_with_alignment_ref_sequence_wrong_length(self, test_file, existing_file, ref_seq, ref_file):
"""Test that including a reference sequence with a length different than the existing alignment fails."""
ref_seq.seq = ref_seq.seq[:-3]
with open(ref_file, "w") as fh:
SeqIO.write(ref_seq, fh, "fasta")
with pytest.raises(align.AlignmentError):
align.prepare([test_file,], existing_file, "out", None, ref_file)
def test_prepare_with_alignment_with_named_ref_missing(
self, test_with_ref, existing_file, ref_seq):
"""We're given a ref_name and an existing alignment, but the ref doesn't exist in the existing alignment."""
with pytest.raises(align.AlignmentError):
align.prepare([
test_with_ref,
], existing_file, "dontcare", ref_seq.id, None)
def test_prepare_no_alignment_with_named_ref_missing(
self, test_file, ref_seq):
"""We're given a ref_name, but it does not exist in the test file"""
with pytest.raises(align.AlignmentError):
align.prepare([
test_file,
], None, "dontcare", ref_seq.id, None)
def test_prepare_with_alignment_with_ref_seq(self, test_file, test_seqs,
existing_file, existing_aln,
ref_seq, ref_file, out_file):
"""Test that, given a set of test sequences, an existing alignment, and a reference sequence, the reference
is added to the existing alignment and no other changes are made."""
aln_outfile, seqs_outfile, ref_name = align.prepare([
test_file,
], existing_file, out_file, None, ref_file)
assert ref_name == ref_seq.id, "Didn't return strain name from refrence file"
assert os.path.isfile(
aln_outfile), "Didn't write existing alignment where it said"
assert aln_outfile != existing_aln, "Unexpectedly overwrote existing alignment"
# Alignment file should have the reference added
aln_output = SeqIO.to_dict(SeqIO.parse(aln_outfile, "fasta"))
assert aln_output[
ref_seq.
id].seq == ref_seq.seq, "Reference sequence not added to alignment"
for seq in existing_aln:
assert seq in aln_output, "Some existing alignment sequences dropped unexpectedly"
assert aln_output[seq].seq == existing_aln[
seq].seq, "Some existing alignment sequences changed unexpectedly"
# test sequences should be unchanged
assert os.path.isfile(
seqs_outfile), "Didn't write test sequences where it said"
seq_output = SeqIO.to_dict(SeqIO.parse(seqs_outfile, "fasta"))
for seq in test_seqs:
assert seq in seq_output, "Some test sequences unexpectedly dropped"
assert seq_output[seq].seq == test_seqs[
seq].seq, "Some test sequences changed unexpectedly"
assert seq_output.keys() == test_seqs.keys()
def test_prepare_with_alignment_with_ref_name(self, test_file, test_seqs,
existing_with_ref,
existing_aln, ref_seq,
out_file):
"""Test that, given a set of test sequences, an existing alignment, and a reference sequence name, no changes are made."""
aln_outfile, seqs_outfile, _ = align.prepare([
test_file,
], existing_with_ref, out_file, ref_seq.id, None)
assert os.path.isfile(
aln_outfile), "Didn't write existing alignment where it said"
assert aln_outfile == existing_with_ref, "Rewrote the alignment file unexpectedly"
# Alignment file should be unchanged
aln_output = SeqIO.to_dict(SeqIO.parse(aln_outfile, "fasta"))
assert aln_output[
ref_seq.
id].seq == ref_seq.seq, "Reference sequence dropped from alignment"
for seq in existing_aln:
assert seq in aln_output, "Some existing alignment sequences dropped unexpectedly"
assert aln_output[seq].seq == existing_aln[
seq].seq, "Some existing alignment sequences changed unexpectedly"
# test sequences should be unchanged
assert os.path.isfile(
seqs_outfile), "Didn't write test sequences where it said"
seq_output = SeqIO.to_dict(SeqIO.parse(seqs_outfile, "fasta"))
for seq in test_seqs:
assert seq in seq_output, "Some test sequences unexpectedly dropped"
assert seq_output[seq].seq == test_seqs[
seq].seq, "Some test sequences changed unexpectedly"
assert seq_output.keys() == test_seqs.keys()
def test_prepare_no_alignment_or_ref(self, test_file, test_seqs, out_file):
_, output, _ = align.prepare([
test_file,
], None, out_file, None, None)
assert os.path.isfile(output), "Didn't write sequences where it said"
for name, seq in SeqIO.to_dict(SeqIO.parse(output, "fasta")).items():
assert seq.seq == test_seqs[name].seq
def test_prepare_no_alignment_with_ref_name(self, test_with_ref, test_seqs, ref_seq, out_file):
_, output_fn, _ = align.prepare([test_with_ref,], None, out_file, ref_seq.id, None)
assert os.path.isfile(output_fn), "Didn't write sequences where it said"
output = SeqIO.to_dict(SeqIO.parse(output_fn, "fasta"))
assert output[ref_seq.id].seq == ref_seq.seq, "Reference sequence was not added to test sequences"
for seq in test_seqs:
assert seq in output, "Some test sequences dropped unexpectedly"
assert output[seq].seq == test_seqs[seq].seq, "Some test sequences changed unexpectedly"
def test_prepare_with_alignment_with_duplicate_sequences(
self, test_file, test_seqs, existing_file, existing_aln, out_file):
"""Test that sequences matching the alignment are removed from the input sequences"""
_, seq_outfile, _ = align.prepare([existing_file, test_file],
existing_file, out_file, None, None)
seq_output = SeqIO.to_dict(SeqIO.parse(seq_outfile, "fasta"))
assert seq_output.keys() == test_seqs.keys(
), "Did not strip duplicate sequences from test input!"
def test_prepare_no_alignment_multiple_test_seqs(self, test_file, test_seqs, ref_file, ref_seq, out_file):
"""Test that we can pass multiple sequence files to prepare() and get one unified file back"""
# bit of a kludge, but gets us the reference strain in the input files
_, seq_outfile, _ = align.prepare([test_file, ref_file], None, out_file, ref_seq.id, None)
seq_output = SeqIO.to_dict(SeqIO.parse(seq_outfile, "fasta"))
assert seq_output.keys() == set(test_seqs.keys()) | {ref_seq.id}, "Did not combine the two files"
assert seq_output[ref_seq.id].seq == ref_seq.seq, "Missing sequence from second file"
for seq in test_seqs:
assert seq in seq_output, "Some test sequences unexpectedly dropped"
assert seq_output[seq].seq == test_seqs[seq].seq, "Some test sequences changed unexpectedly"
def test_prepare_no_alignment_with_ref_file(self, test_file, test_seqs, ref_file, ref_seq, out_file):
_, output_fn, ref_name = align.prepare([test_file,], None, out_file, None, ref_file)
assert ref_name == ref_seq.id, "Didn't return strain name from refrence file"
assert os.path.isfile(output_fn), "Didn't write sequences where it said"
output = list(SeqIO.parse(output_fn, "fasta")) # order matters
assert output[0].id == ref_seq.id, "Reference sequence is not the first sequence in ouput file!"
output_names = {record.name for record in output}
assert all(name in output_names for name in test_seqs), "Some test sequences dropped unexpectedly"
for record in output[1:]:
assert record.seq == test_seqs[record.id].seq, "Some test sequences changed unexpectedly"
