def test_pad_softclip_1(tmpdir):
"it should memorize the result"
make_bam(tmpdir.strpath, """
r1 + __.*.......
r1 - .*.......__
""")
o = Namespace(verbos=False, mismatch_limit=-1)
sam = AlignmentFile(tmpdir.join("test.bam").strpath)
a = pad_softclip(sam)
b = pad_softclip(sam)
assert a is b
def anno(line):
line = line.rstrip().split('\t')
chr, pos, _, ref, alt = line[:5]
line[-3] += ':UDP' # format
line[-2] += ':' # gdna
line[-1] += ':' # cfdna
for i, sam in enumerate((o.cfdna, o.gdna)):
if sam != None:
reads = get_reads(o, sam, chr, pos)
unique_pairs, unique_single, *_ = aggregate_reads(
o, reads, None if o.fast else pad_softclip(sam))
mor, mnr, msr, oor, onr, osr, moa, mna, msa, ooa, ona, osa, _ = count_different_type(
o, unique_pairs, unique_single, alt, ref)
if o.simple:
line[-i - 1] += ','.join(
map(str, (moa, mna + msa, ooa, ona + osa)))
else:
line[-i - 1] += ','.join(
map(str, (mor, mnr, msr, oor, onr, osr, moa, mna,
msa, ooa, ona, osa)))
print(file=fout, sep='\t', *line)
def test_pad_softclip_3(tmpdir):
"it should pad softclipped bases"
make_bam(
tmpdir.strpath, """
123456789_123
r1 + __.*.......
r1 - .*.........
r2 - ...*.......
r2 + .*.......__
""")
o = Namespace(verbos=False, mismatch_limit=-1)
sam = AlignmentFile(tmpdir.join("test.bam").strpath)
adjusted_pos = pad_softclip(sam)
assert adjusted_pos["r1"] == (0, 13) # 0-based position
assert adjusted_pos["r2"] == (0, 13)
def test_pad_softclip_2(tmpdir):
"it should ignore more than two reads which share the same name"
make_bam(
tmpdir.strpath, """
r1 + __.*.......
r1 - .*.......__
r1 - .*.......__
r2 + .*.......__
r2 - .*.......__
""")
o = Namespace(verbos=False, mismatch_limit=-1)
sam = AlignmentFile(tmpdir.join("test.bam").strpath)
adjusted_pos = pad_softclip(sam)
assert sum(1 for startpos, length in adjusted_pos.values()
if startpos != -1) == 1