def test_postprocess_fill_gaps(self, existing_file, existing_aln, ref_seq, fill_gaps):
"""Postprocess should make the gaps ambiguous only if requested"""
align.postprocess(existing_file, None, True, fill_gaps)
output = SeqIO.to_dict(SeqIO.parse(existing_file, "fasta"))
for name, record in output.items():
for idx, site in enumerate(existing_aln[name].seq):
if site == "-":
assert (record.seq[idx] == "N") == fill_gaps
def test_postprocess_prettify_alignment(self, tmpdir, existing_aln, ref_seq):
"""Postprocess should strip _R_ from reverse-complemented strains and convert all sites to uppercase"""
ref_seq.name = "_R_" + ref_seq.name
post_align = write_strains(tmpdir, "post_align", [ref_seq] + list(existing_aln.values()))
align.postprocess(post_align, None, True, False)
output = SeqIO.to_dict(SeqIO.parse(post_align, "fasta"))
for name, record in output.items():
assert record.seq == record.seq.upper(), "Sequence was not made uppercase"
assert "_R_" not in name, "Reverse-complement prefix not removed"
def test_postprocess_strip_non_reference(self, tmpdir, ref_seq, ref_file):
"""Postprocess should strip gaps in the reference sequence from other sequences, but not gaps in those sequences"""
expected_length = len(ref_seq.seq) - ref_seq.seq.count("-")
gapped_seq = MutableSeq(str(ref_seq.seq))
gapped_seq[1] = "-"
gapped = SeqRecord(gapped_seq, "GAP")
gap_file = write_strains(tmpdir, "gaps", [ref_seq, gapped])
align.postprocess(gap_file, ref_seq.id, True, False)
output = SeqIO.to_dict(SeqIO.parse(gap_file, "fasta"))
assert "-" not in output[ref_seq.id].seq
assert output["GAP"].seq.count("-") == 1
for record in output.values():
assert len(record.seq) == expected_length
def test_postprocess_remove_reference(self, existing_with_ref, ref_seq, keep_ref):
"""Postprocess should remove the reference strain only if requested"""
align.postprocess(existing_with_ref, ref_seq.id, keep_ref, False)
output = SeqIO.to_dict(SeqIO.parse(existing_with_ref, "fasta"))
assert (ref_seq.id in output) == keep_ref