def test_downgrade_read_edges_binary(self):
binary = os.path.join(BAM_BIN_DIR, "downgrade_bam_edge_qual")
bam_fpath = os.path.join(TEST_DATA_DIR, "sample_rev.bam")
with NamedTemporaryFile() as out_fhand:
cmd = [binary, "-o", out_fhand.name, bam_fpath]
check_call(cmd)
sam = AlignmentFile(out_fhand.name)
res = [0, 0]
read = sam.next()
assert list(read.query_qualities[:2]) == res
assert read.get_tag("dl") == "8)5B"
assert read.get_tag("dr") == "8?>>"
def test_downgrade_read_edges_binary(self):
binary = os.path.join(BAM_BIN_DIR, 'downgrade_bam_edge_qual')
bam_fpath = os.path.join(TEST_DATA_DIR, 'sample_rev.bam')
with NamedTemporaryFile() as out_fhand:
cmd = [binary, '-o', out_fhand.name, bam_fpath]
check_call(cmd)
sam = AlignmentFile(out_fhand.name)
res = [0, 0]
read = sam.next()
assert list(read.query_qualities[:2]) == res
assert read.get_tag('dl') == '8)5B'
assert read.get_tag('dr') == '8?>>'
def test_downngrade_read_edges(self):
# With softclip
bam_fpath = os.path.join(TEST_DATA_DIR, 'sample.bam')
sam = AlignmentFile(bam_fpath)
aligned_read = sam.next()
_downgrade_edge_qualities(aligned_read, size=4, qual_to_substract=30)
res = [
9, 9, 9, 9, 9, 9, 3, 9, 8, 8, 9, 9, 9, 9, 9, 39, 39, 39, 38, 38,
36, 33, 36, 38, 36, 38, 38, 38, 38, 39, 39, 38, 38, 38, 9, 9, 9, 9
]
assert list(aligned_read.query_qualities) == res
# without softclip
sam = AlignmentFile(os.path.join(TEST_DATA_DIR, 'seqs.bam'))
aligned_read = sam.next()
_downgrade_edge_qualities(aligned_read, size=4, qual_to_substract=30)
expected = [
11, 13, 11, 11, 37, 43, 43, 46, 46, 57, 57, 48, 57, 57, 42, 41, 32,
35, 38, 38, 38, 38, 41, 41, 39, 37, 37, 44, 42, 48, 47, 57, 47, 47,
48, 47, 57, 57, 54, 48, 57, 48, 54, 50, 50, 50, 50, 50, 57, 59, 54,
54, 54, 57, 57, 59, 57, 52, 52, 52, 52, 57, 57, 57, 57, 52, 52, 52,
52, 29, 27, 27, 22
]
assert list(aligned_read.query_qualities) == expected
# reverse
# rev seqs (sam specification puts all the alignment query
# forward(cigar, seq, qual, ...). Reverse is inly noted in the flag
bam_fpath = os.path.join(TEST_DATA_DIR, 'sample_rev.bam')
sam = AlignmentFile(bam_fpath)
aligned_read = sam.next()
aligned_read = sam.next()
aligned_read = sam.next()
_downgrade_edge_qualities(aligned_read, size=4, qual_to_substract=30)
res = [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 3, 0]
assert list(aligned_read.query_qualities[:14]) == res
def test_downgrade_read_edges_binary(self):
binary = os.path.join(BAM_BIN_DIR, 'downgrade_bam_edge_qual')
bam_fpath = os.path.join(TEST_DATA_DIR, 'sample_rev.bam')
with NamedTemporaryFile() as out_fhand:
cmd = [binary, '-o', out_fhand.name, bam_fpath]
check_call(cmd)
sam = AlignmentFile(out_fhand.name)
res = [0, 0]
read = sam.next()
assert list(read.query_qualities[:2]) == res
assert read.get_tag('dl') == '8)5B'
assert read.get_tag('dr') == '8?>>'
# check bam substitution
with TemporaryDir() as tmp_dir:
dirname = tmp_dir.name
shutil.copy(bam_fpath, dirname)
bam_fpath = os.path.join(dirname, os.path.basename(bam_fpath))
cmd = [binary, bam_fpath, '-t', '/home/peio/']
check_call(cmd)
sam = AlignmentFile(bam_fpath)
res = [0, 0]
read = sam.next()
assert list(read.query_qualities[:2]) == res
assert read.get_tag('dl') == '8)5B'
assert read.get_tag('dr') == '8?>>'
# we can not downgrade an already downgraded bam
try:
cmd = [binary, bam_fpath]
stderr_fhand = NamedTemporaryFile()
check_call(cmd, stderr=stderr_fhand)
self.fail('CalledProcessError expected')
except CalledProcessError:
stderr_fhand.flush()
msg = 'RuntimeError: Edge qualities already downgraded'
if msg not in open(stderr_fhand.name).read():
raise
def test_calmd_bam(self):
ref_fpath = os.path.join(TEST_DATA_DIR, "CUUC00007_TC01.fasta")
bam_fpath = os.path.join(TEST_DATA_DIR, "sample.bam")
orig_qual = AlignmentFile(bam_fpath).next().qual
try:
out_bam = NamedTemporaryFile()
calmd_bam(bam_fpath, ref_fpath, out_bam.name)
samfile = AlignmentFile(out_bam.name)
calmd_qual = samfile.next().qual
assert orig_qual != calmd_qual
assert calmd_qual == "HHHHHHBHGGH!!!!!!!!!!!!!!!!!!!!!!!!!!!"
finally:
if os.path.exists(out_bam.name):
out_bam.close()
def test_calmd_bam(self):
ref_fpath = os.path.join(TEST_DATA_DIR, 'CUUC00007_TC01.fasta')
bam_fpath = os.path.join(TEST_DATA_DIR, 'sample.bam')
orig_qual = AlignmentFile(bam_fpath).next().qual
try:
out_bam = NamedTemporaryFile()
calmd_bam(bam_fpath, ref_fpath, out_bam.name)
samfile = AlignmentFile(out_bam.name)
calmd_qual = samfile.next().qual
assert orig_qual != calmd_qual
assert calmd_qual == 'HHHHHHBHGGH!!!!!!!!!!!!!!!!!!!!!!!!!!!'
finally:
if os.path.exists(out_bam.name):
out_bam.close()
def test_downngrade_read_edges(self):
# With softclip
bam_fpath = os.path.join(TEST_DATA_DIR, "sample.bam")
sam = AlignmentFile(bam_fpath)
aligned_read = sam.next()
_downgrade_edge_qualities(aligned_read, size=4, qual_to_substract=30)
res = [
9,
9,
9,
9,
9,
9,
3,
9,
8,
8,
9,
9,
9,
9,
9,
39,
39,
39,
38,
38,
36,
33,
36,
38,
36,
38,
38,
38,
38,
39,
39,
38,
38,
38,
9,
9,
9,
9,
]
assert list(aligned_read.query_qualities) == res
# without softclip
sam = AlignmentFile(os.path.join(TEST_DATA_DIR, "seqs.bam"))
aligned_read = sam.next()
_downgrade_edge_qualities(aligned_read, size=4, qual_to_substract=30)
expected = [
11,
13,
11,
11,
37,
43,
43,
46,
46,
57,
57,
48,
57,
57,
42,
41,
32,
35,
38,
38,
38,
38,
41,
41,
39,
37,
37,
44,
42,
48,
47,
57,
47,
47,
48,
47,
57,
57,
54,
48,
57,
48,
54,
50,
50,
50,
50,
50,
57,
59,
54,
54,
54,
57,
57,
59,
57,
52,
52,
52,
52,
57,
57,
57,
57,
52,
52,
52,
52,
29,
27,
27,
22,
]
assert list(aligned_read.query_qualities) == expected
# reverse
# rev seqs (sam specification puts all the alignment query
# forward(cigar, seq, qual, ...). Reverse is inly noted in the flag
bam_fpath = os.path.join(TEST_DATA_DIR, "sample_rev.bam")
sam = AlignmentFile(bam_fpath)
aligned_read = sam.next()
aligned_read = sam.next()
aligned_read = sam.next()
_downgrade_edge_qualities(aligned_read, size=4, qual_to_substract=30)
res = [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 3, 0]
assert list(aligned_read.query_qualities[:14]) == res
def test_downgrade_read_edges(self):
# With softclip
bam_fpath = os.path.join(TEST_DATA_DIR, 'sample.bam')
sam = AlignmentFile(bam_fpath)
aligned_read = sam.next()
_downgrade_edge_qualities(aligned_read, size=4, qual_to_substract=30)
res = [
9, 9, 9, 9, 9, 9, 3, 9, 8, 8, 9, 9, 9, 9, 9, 39, 39, 39, 38, 38,
36, 33, 36, 38, 36, 38, 38, 38, 38, 39, 39, 38, 38, 38, 9, 9, 9, 9
]
assert list(aligned_read.query_qualities) == res
# without softclip
sam = AlignmentFile(os.path.join(TEST_DATA_DIR, 'seqs.bam'))
aligned_read = sam.next()
_downgrade_edge_qualities(aligned_read, size=4, qual_to_substract=30)
expected = [
11, 13, 11, 11, 37, 43, 43, 46, 46, 57, 57, 48, 57, 57, 42, 41, 32,
35, 38, 38, 38, 38, 41, 41, 39, 37, 37, 44, 42, 48, 47, 57, 47, 47,
48, 47, 57, 57, 54, 48, 57, 48, 54, 50, 50, 50, 50, 50, 57, 59, 54,
54, 54, 57, 57, 59, 57, 52, 52, 52, 52, 57, 57, 57, 57, 52, 52, 52,
52, 29, 27, 27, 22
]
assert list(aligned_read.query_qualities) == expected
# reverse
# rev seqs (sam specification puts all the alignment query
# forward(cigar, seq, qual, ...). Reverse is inly noted in the flag
bam_fpath = os.path.join(TEST_DATA_DIR, 'sample_rev.bam')
sam = AlignmentFile(bam_fpath)
aligned_read = sam.next()
aligned_read = sam.next()
aligned_read = sam.next()
original_qual = aligned_read.query_qualities
_downgrade_edge_qualities(aligned_read, size=4, qual_to_substract=30)
res = [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 3, 0]
assert list(aligned_read.query_qualities[:14]) == res
# check that we can restore the cuals from the tag
_restore_qual_from_tag(aligned_read)
assert original_qual == aligned_read.query_qualities
# only restore left quals
aligned_read = sam.next()
original_qual = aligned_read.query_qualities
_downgrade_edge_qualities(aligned_read, size=4, qual_to_substract=30)
aligned_read.set_tag(RIGTH_DOWNGRADED_TAG, None)
changed_rquals = aligned_read.query_qualities[-5:]
_restore_qual_from_tag(aligned_read)
assert aligned_read.query_qualities[-5:] == changed_rquals
assert aligned_read.query_qualities[:10] == original_qual[:10]
# only restore rigth quals
sam = AlignmentFile(os.path.join(TEST_DATA_DIR, 'seqs.bam'))
aligned_read = sam.next()
original_qual = aligned_read.query_qualities
_downgrade_edge_qualities(aligned_read, size=4, qual_to_substract=30)
aligned_read.set_tag(LEFT_DOWNGRADED_TAG, None)
changed_lquals = aligned_read.query_qualities[:5]
_restore_qual_from_tag(aligned_read)
assert aligned_read.query_qualities[:5] == changed_lquals
assert aligned_read.query_qualities[10:] == original_qual[10:]
