def testMinScoreNoScores(self):
"""
A request for reads with alignment scores no lower than a given value
must produce an empty result when no alignments have scores.
"""
data = '\n'.join([
'@SQ SN:ref1 LN:10',
'query1 0 ref1 2 60 2=2X2M * 0 0 TCTAGG ZZZZZZ',
'query2 0 ref1 2 60 2= * 0 0 TC ZZ',
]).replace(' ', '\t')
with dataFile(data) as filename:
sf = SAMFilter(filename, minScore=6)
self.assertEqual([], list(sf.alignments()))
def testDropSecondary(self):
"""
Dropping matches flagged as secondary must give the expected result.
"""
data = '\n'.join([
'@SQ SN:ref1 LN:10',
'query1 0 ref1 2 60 2=2X2M * 0 0 TCTAGG ZZZZZZ',
'query2 256 ref1 2 60 2= * 0 0 TC ZZ',
]).replace(' ', '\t')
with dataFile(data) as filename:
sf = SAMFilter(filename, dropSecondary=True)
(alignment,) = list(sf.alignments())
self.assertEqual('query1', alignment.query_name)
def testAlignmentCount(self):
"""
When all queries have been yielded, the alignment count must be
as expected.
"""
data = '\n'.join([
'@SQ SN:ref1 LN:10',
'query1 0 ref1 2 60 2=2X2M * 0 0 TCTAGG 123456',
'query2 0 ref1 2 60 2= * 0 0 TC XY',
]).replace(' ', '\t')
with dataFile(data) as filename:
sf = SAMFilter(filename)
list(sf.alignments())
self.assertEqual(2, sf.alignmentCount)
def testStoreQueryIds(self):
"""
If we request that query ids are saved, they must be.
"""
data = '\n'.join([
'@SQ SN:ref1 LN:10',
'query1 0 ref1 2 60 2=2X2M * 0 0 TCTAGG 123456',
'query2 0 ref1 2 60 2= * 0 0 TC XY',
'query2 0 ref1 2 60 2= * 0 0 TC XY',
]).replace(' ', '\t')
with dataFile(data) as filename:
sf = SAMFilter(filename, storeQueryIds=True)
list(sf.alignments())
self.assertEqual({'query1', 'query2'}, sf.queryIds)
def testMinLength(self):
"""
A request for reads that are only longer than a certain value should
result in the expected result.
"""
data = '\n'.join([
'@SQ SN:ref1 LN:10',
'query1 0 ref1 2 60 2=2X2M * 0 0 TCTAGG ZZZZZZ',
'query2 0 ref1 2 60 2= * 0 0 TC ZZ',
]).replace(' ', '\t')
with dataFile(data) as filename:
filterRead = ReadFilter(minLength=6).filter
sf = SAMFilter(filename, filterRead=filterRead)
(alignment,) = list(sf.alignments())
self.assertEqual('query1', alignment.query_name)
def testNoQuality(self):
"""
If an alignment has * for the quality string, the filter must
return an alignment with a C{None} quality value.
"""
data = '\n'.join([
'@SQ SN:ref LN:10',
'query1 4 * 0 0 6M * 0 0 TCTAGG *',
]).replace(' ', '\t')
with dataFile(data) as filename:
sf = SAMFilter(filename)
(alignment,) = list(sf.alignments())
self.assertEqual('query1', alignment.query_name)
self.assertEqual('TCTAGG', alignment.query_sequence)
self.assertIsNone(alignment.query_qualities)
def testMaxScore(self):
"""
A request for reads with alignment scores no higher than a given value
must produce the expected result when some alignments have scores.
"""
data = '\n'.join([
'@SQ SN:ref1 LN:10',
'query1 0 ref1 2 60 2=2X2M * 0 0 TCTAGG ZZZZZZ AS:i:10',
'query2 0 ref1 2 60 2= * 0 0 TC ZZ',
'query3 0 ref1 2 60 2=2X2M * 0 0 TCTAGG ZZZZZZ AS:i:3',
]).replace(' ', '\t')
with dataFile(data) as filename:
sf = SAMFilter(filename, maxScore=6)
(alignment,) = list(sf.alignments())
self.assertEqual('query3', alignment.query_name)
def testKeepQualityControlFailures(self):
"""
Keeping matches flagged as quality control failures must give the
expected result.
"""
data = '\n'.join([
'@SQ SN:ref1 LN:10',
'query1 0 ref1 2 60 2=2X2M * 0 0 TCTAGG ZZZZZZ',
'query2 512 ref1 4 60 2= * 0 0 TC ZZ',
]).replace(' ', '\t')
with dataFile(data) as filename:
sf = SAMFilter(filename, keepQCFailures=True)
(alignment1, alignment2) = list(sf.alignments())
self.assertEqual('query1', alignment1.query_name)
self.assertEqual('query2', alignment2.query_name)
def testCloseButNoCIGAR(self):
"""
An unmapped query with no CIGAR string must be passed through
unchanged if dropUnmapped is not specified.
"""
data = '\n'.join([
'@SQ SN:ref LN:10',
'query1 4 * 0 0 * * 0 0 TCTAGG ZZZZZZ',
]).replace(' ', '\t')
with dataFile(data) as filename:
sf = SAMFilter(filename)
(alignment,) = list(sf.alignments())
self.assertEqual('query1', alignment.query_name)
self.assertEqual('TCTAGG', alignment.query_sequence)
self.assertEqual('ZZZZZZ', ''.join(
map(lambda x: chr(x + 33), alignment.query_qualities)))
def testMinAndMaxScore(self):
"""
A request for reads with alignment scores no lower or higher than
given values must produce the expected result.
"""
data = '\n'.join([
'@SQ SN:ref1 LN:10',
'query1 0 ref1 2 60 2=2X2M * 0 0 TCTAGG ZZZZZZ AS:i:10',
'query2 0 ref1 2 60 2=2X2M * 0 0 TCTAGG ZZZZZZ AS:i:12',
'query3 0 ref1 2 60 2= * 0 0 TC ZZ',
'query4 0 ref1 2 60 2=2X2M * 0 0 TCTAGG ZZZZZZ AS:i:3',
'query5 0 ref1 2 60 2=2X2M * 0 0 TCTAGG ZZZZZZ AS:i:2',
]).replace(' ', '\t')
with dataFile(data) as filename:
sf = SAMFilter(filename, minScore=3, maxScore=10)
(alignment1, alignment2) = list(sf.alignments())
self.assertEqual('query1', alignment1.query_name)
self.assertEqual('query4', alignment2.query_name)
