def test_error_open_bamfile(self):
"""
Provide onyl file with no content - error shoud be raised.
"""
bam_fname = get_temp_file_name()
message = r"Error opening BAM file: .*"
with self.assertRaisesRegex(ValueError, message):
xlsites._processs_bam_file(bam_fname, self.metrics, 50, self.tmp)
def test_no_nh_tag(self):
data_no_nh = {
'chromosomes': [('chr1', 3000)],
'segments': [
# No NH tag is set
('name5', 0, 0, 0, 50, [(0, 0)], {})
]
}
bam_fname = make_bam_file(data_no_nh)
message = r'"NH" tag not set for record: .*'
with self.assertRaisesRegex(ValueError, message):
xlsites._processs_bam_file(bam_fname, self.metrics, 10, self.tmp)
def test_low_quality(self):
"""
Unmapped read (FLAG=4):
"""
bam_fname = make_bam_file({
'chromosomes': [('chr1', 3000)],
'segments': [('name1', 0, 0, 0, 3, [(0, 0)], {})],
})
self.metrics.lowmapq_recs = 0
self.metrics.used_recs = 0
xlsites._processs_bam_file(bam_fname, self.metrics, 10, self.tmp)
self.assertEqual(self.metrics.lowmapq_recs, 1)
self.assertEqual(self.metrics.used_recs, 0)
def test_pos_neg_strand(self):
"""
positive, negative strnad
len %2 == 0 na negativnem stnadu
and not %2
"""
bam_fname = make_bam_file({
'chromosomes': [('chr1', 3000)],
'segments': [
# (qname, flag, refname, pos, mapq, cigar, tags)
('_:rbc:AAA', 16, 0, 50, 255, [(0, 100)], {
'NH': 1
}),
('_:rbc:CCC', 0, 0, 50, 255, [(0, 101)], {
'NH': 1
}),
],
})
grouped = xlsites._processs_bam_file(bam_fname, self.metrics, 10,
self.tmp)
expected = {
('chr1', '-'): {
150: {
'AAA': [(99, 50, 100, 1, 0)]
}
},
('chr1', '+'): {
49: {
'CCC': [(100, 150, 101, 1, 0)]
}
},
}
self.assertEqual(grouped, expected)
def test_diff_barcodes(self):
"""
Different barcodes on same position.
"""
bam_fname = make_bam_file({
'chromosomes': [('chr1', 3000)],
'segments': [
# (qname, flag, refname, pos, mapq, cigar, tags)
('_:rbc:AAA', 0, 0, 50, 255, [(0, 101)], {'NH': 1}),
('_:rbc:CCC', 0, 0, 50, 255, [(0, 101)], {'NH': 1}),
('_:rbc:CCC', 0, 0, 50, 255, [(0, 101)], {'NH': 1}),
('_:rbc:GGG', 0, 0, 50, 255, [(0, 101)], {'NH': 1}),
],
}, rnd_seed=0)
grouped = list(xlsites._processs_bam_file(bam_fname, self.metrics, 10, self.tmp))
expected = [
(('chr1', '+'), 0.0167, {
49: {
'AAA': [(100, 150, 101, 1, 0)],
'CCC': [(100, 150, 101, 1, 0), (100, 150, 101, 1, 0)],
'GGG': [(100, 150, 101, 1, 0)],
}
}),
]
self.assertEqual(grouped, expected)
def test_unmapped(self):
"""
Unmapped read (FLAG=4):
"""
bam_fname = make_bam_file({
'chromosomes': [('chr1', 3000)],
'segments': [('name1', 4, 0, 0, 0, [(0, 0)], {})],
}, rnd_seed=0)
self.metrics.all_recs = 0
self.metrics.notmapped_recs = 0
self.metrics.used_recs = 0
list(xlsites._processs_bam_file(bam_fname, self.metrics, 0, self.tmp))
self.assertEqual(self.metrics.notmapped_recs, 1)
self.assertEqual(self.metrics.all_recs, 1)
self.assertEqual(self.metrics.used_recs, 0)