def test_map_reads(self):
"""test map_reads"""
reads1 = os.path.join(data_dir, "reads.1.fq")
reads2 = os.path.join(data_dir, "reads.2.fq")
ref_fasta = os.path.join(data_dir, "ref.fa")
tmp_sam = "tmp.test_map_reads.sam"
if os.path.exists(tmp_sam):
os.unlink(tmp_sam)
read_map.map_reads(ref_fasta,
reads1,
reads2,
tmp_sam,
read_group=("1", "GROUP_NAME"))
self.assertTrue(os.path.exists(tmp_sam))
tmp_stats = tmp_sam + ".stats"
expected_stats = os.path.join(data_dir, "flagstat")
utils.syscall("samtools flagstat " + tmp_sam +
" | grep -v primary > " + tmp_stats)
self.assertTrue(filecmp.cmp(expected_stats, tmp_stats, shallow=False))
found_rg_line = False
with open(tmp_sam) as f:
for line in f:
if line == "@RG\tLB:LIB\tID:1\tSM:GROUP_NAME\n":
found_rg_line = True
break
self.assertTrue(found_rg_line)
os.unlink(tmp_sam)
os.unlink(tmp_stats)
def test_map_reads(self):
'''test map_reads'''
reads1 = os.path.join(data_dir, 'reads.1.fq')
reads2 = os.path.join(data_dir, 'reads.2.fq')
ref_fasta = os.path.join(data_dir, 'ref.fa')
tmp_sam = 'tmp.test_map_reads.sam'
if os.path.exists(tmp_sam):
os.unlink(tmp_sam)
read_map.map_reads(ref_fasta,
reads1,
reads2,
tmp_sam,
read_group=('1', 'GROUP_NAME'))
self.assertTrue(os.path.exists(tmp_sam))
tmp_stats = tmp_sam + '.stats'
expected_stats = os.path.join(data_dir, 'flagstat')
utils.syscall('samtools flagstat ' + tmp_sam + ' > ' + tmp_stats)
self.assertTrue(filecmp.cmp(expected_stats, tmp_stats, shallow=False))
found_rg_line = False
with open(tmp_sam) as f:
for line in f:
if line == '@RG\tLB:LIB\tID:1\tSM:GROUP_NAME\n':
found_rg_line = True
break
self.assertTrue(found_rg_line)
os.unlink(tmp_sam)
os.unlink(tmp_stats)
def test_map_reads_markdup_and_rmdup(self):
"""test map_reads rmdup and markdup"""
with self.assertRaises(Exception):
read_map.map_reads("ref_fasta",
"reads1",
"reads2",
"sam",
rmdup=True,
markdup=True)
def test_map_reads_markdup_and_rmdup(self):
'''test map_reads rmdup and markdup'''
with self.assertRaises(read_map.Error):
read_map.map_reads('ref_fasta',
'reads1',
'reads2',
'sam',
rmdup=True,
markdup=True)
def test_map_reads_secondary_hits_removed(self):
"""test map_reads secondary hits get removed"""
reads1 = os.path.join(data_dir, "secondary_hits_removed.reads_1.fq")
reads2 = os.path.join(data_dir, "secondary_hits_removed.reads_2.fq")
ref_fasta = os.path.join(data_dir, "secondary_hits_removed.ref.fa")
tmp_sam = "tmp.test_map_reads.sam"
if os.path.exists(tmp_sam):
os.unlink(tmp_sam)
read_map.map_reads(ref_fasta, reads1, reads2, tmp_sam)
# bwa mem reports one secondary alignment, so the 1 read pair makes
# 3 SAM records. So should have 2 records after removing secondary match.
self.assertEqual(2, utils.sam_record_count(tmp_sam))
os.unlink(tmp_sam)
def test_map_reads_markdup(self):
'''test map_reads markdup'''
reads1 = os.path.join(data_dir, 'reads.1.fq')
reads2 = os.path.join(data_dir, 'reads.2.fq')
ref_fasta = os.path.join(data_dir, 'ref.fa')
tmp_sam = 'tmp.test_map_reads.sam'
if os.path.exists(tmp_sam):
os.unlink(tmp_sam)
read_map.map_reads(ref_fasta, reads1, reads2, tmp_sam, markdup=True)
self.assertTrue(os.path.exists(tmp_sam))
tmp_stats = tmp_sam + '.stats'
expected_stats = os.path.join(data_dir, 'markdup.flagstat')
utils.syscall('samtools flagstat ' + tmp_sam + ' > ' + tmp_stats)
self.assertTrue(filecmp.cmp(expected_stats, tmp_stats, shallow=False))
os.unlink(tmp_sam)
os.unlink(tmp_stats)
def test_map_reads_markdup(self):
"""test map_reads markdup"""
reads1 = os.path.join(data_dir, "reads.1.fq")
reads2 = os.path.join(data_dir, "reads.2.fq")
ref_fasta = os.path.join(data_dir, "ref.fa")
tmp_sam = "tmp.test_map_reads.sam"
if os.path.exists(tmp_sam):
os.unlink(tmp_sam)
read_map.map_reads(ref_fasta, reads1, reads2, tmp_sam, markdup=True)
self.assertTrue(os.path.exists(tmp_sam))
tmp_stats = tmp_sam + ".stats"
expected_stats = os.path.join(data_dir, "markdup.flagstat")
utils.syscall("samtools flagstat " + tmp_sam + " > " + tmp_stats)
self.assertTrue(filecmp.cmp(expected_stats, tmp_stats, shallow=False))
os.unlink(tmp_sam)
os.unlink(tmp_stats)
def _map_reads(cls, ref_fasta, reads1, reads2, outfile):
read_map.map_reads(ref_fasta, reads1, reads2, outfile, markdup=True)
