def testReadSetFilterAllowAnything(self):
"""
The filter function must work correctly when passed a 0.0 value for
minNewReads, i.e. that considers any read set sufficiently novel.
"""
mockOpener = mock_open(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n' + dumps(RECORD2) + '\n' +
dumps(RECORD3) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
reads.add(Read('id2', 'A' * 70))
reads.add(Read('id3', 'A' * 70))
readsAlignments = DiamondReadsAlignments(reads, 'file.json')
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.filter(minNewReads=0.0)
self.assertEqual(
[
'gi|887699|gb|DQ37780 Cowpox virus 15',
'gi|887699|gb|DQ37780 Monkeypox virus 456',
'gi|887699|gb|DQ37780 Mummypox virus 3000 B.C.',
'gi|887699|gb|DQ37780 Squirrelpox virus 1296/99',
'gi|887699|gb|DQ37780 Squirrelpox virus 55',
],
sorted(result))
def testWithScoreBetterThan_EValue(self):
"""
The filter function work correctly when passed a value for
withScoreBetterThan when using e values.
"""
mockOpener = mockOpen(
read_data=(dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n' + dumps(RECORD2) + '\n' +
dumps(RECORD3) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
reads.add(Read('id2', 'A' * 70))
reads.add(Read('id3', 'A' * 70))
readsAlignments = DiamondReadsAlignments(
reads,
'file.json',
'database.fasta',
scoreClass=LowerIsBetterScore)
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.filter(withScoreBetterThan=1e-10)
self.assertEqual([
'gi|887699|gb|DQ37780 Squirrelpox virus 1296/99',
], list(result.keys()))
def testMinMedianScore_EValue(self):
"""
The filter function work correctly when passed a value for
minMedianScore when using e values.
"""
mockOpener = mockOpen(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n' + dumps(RECORD2) + '\n' +
dumps(RECORD3) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
reads.add(Read('id2', 'A' * 70))
reads.add(Read('id3', 'A' * 70))
readsAlignments = BlastReadsAlignments(
reads, 'file.json', scoreClass=LowerIsBetterScore)
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.filter(minMedianScore=1e-9)
self.assertEqual(
[
'gi|887699|gb|DQ37780 Squirrelpox virus 1296/99',
'gi|887699|gb|DQ37780 Squirrelpox virus 55',
],
sorted(result.keys()))
def testCoverageIncludesAll(self):
"""
The coverage function must return an titlesAlignments instance with
all titles if all its titles has sufficient coverage.
"""
mockOpener = mock_open(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n' + dumps(RECORD2) + '\n' +
dumps(RECORD3) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
reads.add(Read('id2', 'A' * 70))
reads.add(Read('id3', 'A' * 70))
readsAlignments = DiamondReadsAlignments(reads, 'file.json')
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.filter(minCoverage=0.0)
self.assertEqual(
[
'gi|887699|gb|DQ37780 Cowpox virus 15',
'gi|887699|gb|DQ37780 Monkeypox virus 456',
'gi|887699|gb|DQ37780 Mummypox virus 3000 B.C.',
'gi|887699|gb|DQ37780 Squirrelpox virus 1296/99',
'gi|887699|gb|DQ37780 Squirrelpox virus 55',
],
sorted(result))
def testFilterWithNoArguments(self):
"""
The filter function must return a TitlesAlignments instance with all
the titles of the original when called with no arguments.
"""
mockOpener = mockOpen(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n' + dumps(RECORD2) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
reads.add(Read('id2', 'A' * 70))
readsAlignments = BlastReadsAlignments(reads, 'file.json')
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.filter()
self.assertEqual(
[
'gi|887699|gb|DQ37780 Cowpox virus 15',
'gi|887699|gb|DQ37780 Monkeypox virus 456',
'gi|887699|gb|DQ37780 Mummypox virus 3000 B.C.',
'gi|887699|gb|DQ37780 Squirrelpox virus 1296/99',
'gi|887699|gb|DQ37780 Squirrelpox virus 55',
],
sorted(result.keys()))
def testCoverageIncludesAll(self):
"""
The coverage function must return an titlesAlignments instance with
all titles if all its titles has sufficient coverage.
"""
mockOpener = mockOpen(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n' + dumps(RECORD2) + '\n' +
dumps(RECORD3) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
reads.add(Read('id2', 'A' * 70))
reads.add(Read('id3', 'A' * 70))
readsAlignments = BlastReadsAlignments(reads, 'file.json')
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.filter(minCoverage=0.0)
self.assertEqual(
[
'gi|887699|gb|DQ37780 Cowpox virus 15',
'gi|887699|gb|DQ37780 Monkeypox virus 456',
'gi|887699|gb|DQ37780 Mummypox virus 3000 B.C.',
'gi|887699|gb|DQ37780 Squirrelpox virus 1296/99',
'gi|887699|gb|DQ37780 Squirrelpox virus 55',
],
sorted(result.keys()))
def testLength(self):
"""
Sorting on sequence length must work, including a secondary sort on
title.
"""
mockOpener = mockOpen(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n' + dumps(RECORD2) + '\n' +
dumps(RECORD3) + '\n'))
with patch('__builtin__.open', mockOpener, create=True):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
reads.add(Read('id2', 'A' * 70))
reads.add(Read('id3', 'A' * 70))
readsAlignments = BlastReadsAlignments(
reads, 'file.json', scoreClass=LowerIsBetterScore)
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.sortTitles('length')
self.assertEqual([
'gi|887699|gb|DQ37780 Squirrelpox virus 55', # 38000
'gi|887699|gb|DQ37780 Squirrelpox virus 1296/99', # 37000
'gi|887699|gb|DQ37780 Monkeypox virus 456', # 35000
'gi|887699|gb|DQ37780 Mummypox virus 3000 B.C.', # 35000
'gi|887699|gb|DQ37780 Cowpox virus 15', # 30000
], result)
def testMedianScore_Bits(self):
"""
Sorting on median score must work when scores are bit scores,
including a secondary sort on title.
"""
mockOpener = mock_open(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n' + dumps(RECORD2) + '\n' +
dumps(RECORD3) + '\n' + dumps(RECORD4) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
reads.add(Read('id2', 'A' * 70))
reads.add(Read('id3', 'A' * 70))
reads.add(Read('id4', 'A' * 70))
readsAlignments = DiamondReadsAlignments(reads, 'file.json')
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.sortTitles('medianScore')
self.assertEqual([
'gi|887699|gb|DQ37780 Squirrelpox virus 55', # 25
'gi|887699|gb|DQ37780 Monkeypox virus 456', # 20
'gi|887699|gb|DQ37780 Mummypox virus 3000 B.C.', # 20
'gi|887699|gb|DQ37780 Squirrelpox virus 1296/99', # 20
'gi|887699|gb|DQ37780 Cowpox virus 15', # 20
], result)
def testTitle(self):
"""
Sorting on title must work.
"""
mockOpener = mockOpen(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n' + dumps(RECORD2) + '\n' +
dumps(RECORD3) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
reads.add(Read('id2', 'A' * 70))
reads.add(Read('id3', 'A' * 70))
readsAlignments = BlastReadsAlignments(
reads, 'file.json', scoreClass=LowerIsBetterScore)
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.sortTitles('title')
self.assertEqual([
'gi|887699|gb|DQ37780 Cowpox virus 15',
'gi|887699|gb|DQ37780 Monkeypox virus 456',
'gi|887699|gb|DQ37780 Mummypox virus 3000 B.C.',
'gi|887699|gb|DQ37780 Squirrelpox virus 1296/99',
'gi|887699|gb|DQ37780 Squirrelpox virus 55',
], result)
def testLength(self):
"""
Sorting on sequence length must work, including a secondary sort on
title.
"""
mockOpener = mockOpen(
read_data=(dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n' + dumps(RECORD2) + '\n' +
dumps(RECORD3) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
reads.add(Read('id2', 'A' * 70))
reads.add(Read('id3', 'A' * 70))
readsAlignments = BlastReadsAlignments(
reads, 'file.json', scoreClass=LowerIsBetterScore)
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.sortTitles('length')
self.assertEqual(
[
'gi|887699|gb|DQ37780 Squirrelpox virus 55', # 38000
'gi|887699|gb|DQ37780 Squirrelpox virus 1296/99', # 37000
'gi|887699|gb|DQ37780 Monkeypox virus 456', # 35000
'gi|887699|gb|DQ37780 Mummypox virus 3000 B.C.', # 35000
'gi|887699|gb|DQ37780 Cowpox virus 15', # 30000
],
result)
def testReadSetFilterAllowAnything(self):
"""
The filter function work correctly when passed a 0.0 value for
minNewReads, i.e. that considers any read set sufficiently novel.
"""
mockOpener = mockOpen(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n' + dumps(RECORD2) + '\n' +
dumps(RECORD3) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
reads.add(Read('id2', 'A' * 70))
reads.add(Read('id3', 'A' * 70))
readsAlignments = BlastReadsAlignments(reads, 'file.json')
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.filter(minNewReads=0.0)
self.assertEqual(
[
'gi|887699|gb|DQ37780 Cowpox virus 15',
'gi|887699|gb|DQ37780 Monkeypox virus 456',
'gi|887699|gb|DQ37780 Mummypox virus 3000 B.C.',
'gi|887699|gb|DQ37780 Squirrelpox virus 1296/99',
'gi|887699|gb|DQ37780 Squirrelpox virus 55',
],
sorted(result.keys()))
def testMaxScore_EValue(self):
"""
Sorting on max score must work when scores are e values, including a
secondary sort on title.
"""
mockOpener = mock_open(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n' + dumps(RECORD2) + '\n' +
dumps(RECORD3) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
reads.add(Read('id2', 'A' * 70))
reads.add(Read('id3', 'A' * 70))
readsAlignments = DiamondReadsAlignments(
reads, 'file.json', scoreClass=LowerIsBetterScore)
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.sortTitles('maxScore')
# self.assertEqual([
# 'gi|887699|gb|DQ37780 Cowpox virus 15', # 1e-6
# 'gi|887699|gb|DQ37780 Mummypox virus 3000 B.C.', # 1e-7
# 'gi|887699|gb|DQ37780 Monkeypox virus 456', # 1e-8
# 'gi|887699|gb|DQ37780 Squirrelpox virus 55', # 1e-10
# 'gi|887699|gb|DQ37780 Squirrelpox virus 1296/99', # 1e-11
# ], result)
self.assertEqual([
'gi|887699|gb|DQ37780 Squirrelpox virus 1296/99', # 1e-11
'gi|887699|gb|DQ37780 Squirrelpox virus 55', # 1e-10
'gi|887699|gb|DQ37780 Monkeypox virus 456', # 1e-8
'gi|887699|gb|DQ37780 Mummypox virus 3000 B.C.', # 1e-7
'gi|887699|gb|DQ37780 Cowpox virus 15', # 1e-6
], result)
def testMaxMatchingReads(self):
"""
The filter function must work correctly when passed a value for
maxMatchingReads.
"""
mockOpener = mock_open(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n' + dumps(RECORD2) + '\n' +
dumps(RECORD3) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
reads.add(Read('id2', 'A' * 70))
reads.add(Read('id3', 'A' * 70))
readsAlignments = DiamondReadsAlignments(reads, 'file.json')
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.filter(maxMatchingReads=1)
# Cowpox virus 15 is not in the results as it is matched by two
# reads.
self.assertEqual(
sorted([
'gi|887699|gb|DQ37780 Squirrelpox virus 1296/99',
'gi|887699|gb|DQ37780 Squirrelpox virus 55',
'gi|887699|gb|DQ37780 Monkeypox virus 456',
'gi|887699|gb|DQ37780 Mummypox virus 3000 B.C.'
]),
sorted(result))
def testReadSetFilterStrict(self):
"""
The filter function work correctly when passed a 1.0 value for
minNewReads.
"""
mockOpener = mockOpen(read_data=(PARAMS + RECORD0 + RECORD1 + RECORD2 +
RECORD3))
with patch.object(builtins, 'open', mockOpener):
readsAlignments = LightReadsAlignments('file.json', DB)
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.filter(minNewReads=1.0)
# Either MUMMYPOX.id invalidates MONKEYPOX.id or vice-versa. It
# depends on Python's dict walking order. Check for both,
# making sure just one of them is true.
assertionCount = 0
if MUMMYPOX.id in result:
self.assertTrue(MONKEYPOX.id in
result.readSetFilter.invalidates(MUMMYPOX.id))
assertionCount += 1
if MONKEYPOX.id in result:
self.assertTrue(MUMMYPOX.id in
result.readSetFilter.invalidates(MONKEYPOX.id))
assertionCount += 1
self.assertEqual(1, assertionCount)
def testTabSeparatedSummary(self):
"""
The summary function must return the correct result.
"""
mockOpener = mock_open(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
readsAlignments = DiamondReadsAlignments(reads, 'f.json')
titlesAlignments = TitlesAlignments(readsAlignments)
summary = titlesAlignments.tabSeparatedSummary(sortOn='title')
expected = (
'0.000297\t'
'20.000000\t'
'20.000000\t'
'1\t'
'1\t'
'37000\t'
'gi|887699|gb|DQ37780 Squirrelpox virus 1296/99'
'\n'
'0.000289\t'
'25.000000\t'
'25.000000\t'
'1\t'
'1\t'
'38000\t'
'gi|887699|gb|DQ37780 Squirrelpox virus 55')
self.assertEqual(expected, summary)
def testFilterWithNoArguments(self):
"""
The filter function must return a TitlesAlignments instance with all
the titles of the original when called with no arguments.
"""
mockOpener = mock_open(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n' + dumps(RECORD2) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
reads.add(Read('id2', 'A' * 70))
readsAlignments = DiamondReadsAlignments(reads, 'file.json')
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.filter()
self.assertEqual(
[
'gi|887699|gb|DQ37780 Cowpox virus 15',
'gi|887699|gb|DQ37780 Monkeypox virus 456',
'gi|887699|gb|DQ37780 Mummypox virus 3000 B.C.',
'gi|887699|gb|DQ37780 Squirrelpox virus 1296/99',
'gi|887699|gb|DQ37780 Squirrelpox virus 55',
],
sorted(result))
def testTitle(self):
"""
Sorting on title must work.
"""
mockOpener = mock_open(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n' + dumps(RECORD2) + '\n' +
dumps(RECORD3) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
reads.add(Read('id2', 'A' * 70))
reads.add(Read('id3', 'A' * 70))
readsAlignments = DiamondReadsAlignments(
reads, 'file.json', scoreClass=LowerIsBetterScore)
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.sortTitles('title')
self.assertEqual([
'gi|887699|gb|DQ37780 Cowpox virus 15',
'gi|887699|gb|DQ37780 Monkeypox virus 456',
'gi|887699|gb|DQ37780 Mummypox virus 3000 B.C.',
'gi|887699|gb|DQ37780 Squirrelpox virus 1296/99',
'gi|887699|gb|DQ37780 Squirrelpox virus 55',
], result)
def testCoverageIncludesSome(self):
"""
The coverage function must return an titlesAlignments instance with
only the expected titles if only some of its titles have sufficient
coverage.
"""
mockOpener = mockOpen(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n' + dumps(RECORD2) + '\n' +
dumps(RECORD3) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
reads.add(Read('id2', 'A' * 70))
reads.add(Read('id3', 'A' * 70))
readsAlignments = BlastReadsAlignments(reads, 'file.json')
titlesAlignments = TitlesAlignments(readsAlignments)
# To understand why the following produces the result it does,
# you need to look at the HSP coverage in sample_data.py and
# calculate the coverage by hand.
result = titlesAlignments.filter(minCoverage=0.0011)
self.assertEqual(
[
'gi|887699|gb|DQ37780 Cowpox virus 15',
'gi|887699|gb|DQ37780 Monkeypox virus 456',
'gi|887699|gb|DQ37780 Mummypox virus 3000 B.C.',
],
sorted(result.keys()))
def testCoverageIncludesSome(self):
"""
The coverage function must return an titlesAlignments instance with
only the expected titles if only some of its titles have sufficient
coverage.
"""
mockOpener = mock_open(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n' + dumps(RECORD2) + '\n' +
dumps(RECORD3) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
reads.add(Read('id2', 'A' * 70))
reads.add(Read('id3', 'A' * 70))
readsAlignments = DiamondReadsAlignments(reads, 'file.json')
titlesAlignments = TitlesAlignments(readsAlignments)
# To understand why the following produces the result it does,
# you need to look at the HSP coverage in sample_data.py and
# calculate the coverage by hand.
result = titlesAlignments.filter(minCoverage=0.0003)
self.assertEqual(
[
'gi|887699|gb|DQ37780 Cowpox virus 15',
'gi|887699|gb|DQ37780 Monkeypox virus 456',
'gi|887699|gb|DQ37780 Mummypox virus 3000 B.C.',
],
sorted(result))
def testMinMedianScore_EValue(self):
"""
The filter function must work correctly when passed a value for
minMedianScore when using e values.
"""
mockOpener = mock_open(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n' + dumps(RECORD2) + '\n' +
dumps(RECORD3) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
reads.add(Read('id2', 'A' * 70))
reads.add(Read('id3', 'A' * 70))
readsAlignments = DiamondReadsAlignments(
reads, 'file.json', scoreClass=LowerIsBetterScore)
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.filter(minMedianScore=1e-9)
self.assertEqual(
[
'gi|887699|gb|DQ37780 Squirrelpox virus 1296/99',
'gi|887699|gb|DQ37780 Squirrelpox virus 55',
],
sorted(result))
def testMedianScore_Bits(self):
"""
Sorting on median score must work when scores are bit scores,
including a secondary sort on title.
"""
mockOpener = mockOpen(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n' + dumps(RECORD2) + '\n' +
dumps(RECORD3) + '\n' + dumps(RECORD4) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
reads.add(Read('id2', 'A' * 70))
reads.add(Read('id3', 'A' * 70))
reads.add(Read('id4', 'A' * 70))
readsAlignments = BlastReadsAlignments(reads, 'file.json')
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.sortTitles('medianScore')
self.assertEqual([
'gi|887699|gb|DQ37780 Squirrelpox virus 55', # 25
'gi|887699|gb|DQ37780 Monkeypox virus 456', # 20
'gi|887699|gb|DQ37780 Mummypox virus 3000 B.C.', # 20
'gi|887699|gb|DQ37780 Squirrelpox virus 1296/99', # 20
'gi|887699|gb|DQ37780 Cowpox virus 15', # 20
], result)
def testMedianScore_EValue(self):
"""
Sorting on median score must work when scores are bit scores,
including a secondary sort on title.
"""
mockOpener = mockOpen(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n' + dumps(RECORD2) + '\n' +
dumps(RECORD3) + '\n' + dumps(RECORD4) + '\n'))
with patch('__builtin__.open', mockOpener, create=True):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
reads.add(Read('id2', 'A' * 70))
reads.add(Read('id3', 'A' * 70))
reads.add(Read('id4', 'A' * 70))
readsAlignments = BlastReadsAlignments(
reads, 'file.json', scoreClass=LowerIsBetterScore)
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.sortTitles('medianScore')
self.assertEqual([
'gi|887699|gb|DQ37780 Squirrelpox virus 1296/99', # 1e-11
'gi|887699|gb|DQ37780 Squirrelpox virus 55', # 1e-10
'gi|887699|gb|DQ37780 Monkeypox virus 456', # 1e-8
'gi|887699|gb|DQ37780 Mummypox virus 3000 B.C.', # 1e-7
'gi|887699|gb|DQ37780 Cowpox virus 15', # worst :-)
], result)
def testMaxTitlesTwoSortOnLength(self):
"""
The filter function must return the two titles whose sequences are the
longest when maxTitles is 2 and sortOn is 'length'.
"""
mockOpener = mockOpen(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n' + dumps(RECORD2) + '\n' +
dumps(RECORD3) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
reads.add(Read('id2', 'A' * 70))
reads.add(Read('id3', 'A' * 70))
readsAlignments = DiamondReadsAlignments(reads, 'file.json',
'database.fasta')
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.filter(maxTitles=2, sortOn='length')
self.assertEqual(
[
'gi|887699|gb|DQ37780 Squirrelpox virus 1296/99',
'gi|887699|gb|DQ37780 Squirrelpox virus 55',
],
sorted(result.keys()))
def testMaxScore_EValue(self):
"""
Sorting on max score must work when scores are e values, including a
secondary sort on title.
"""
mockOpener = mockOpen(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n' + dumps(RECORD2) + '\n' +
dumps(RECORD3) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
reads.add(Read('id2', 'A' * 70))
reads.add(Read('id3', 'A' * 70))
readsAlignments = BlastReadsAlignments(
reads, 'file.json', scoreClass=LowerIsBetterScore)
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.sortTitles('maxScore')
# self.assertEqual([
# 'gi|887699|gb|DQ37780 Cowpox virus 15', # 1e-6
# 'gi|887699|gb|DQ37780 Mummypox virus 3000 B.C.', # 1e-7
# 'gi|887699|gb|DQ37780 Monkeypox virus 456', # 1e-8
# 'gi|887699|gb|DQ37780 Squirrelpox virus 55', # 1e-10
# 'gi|887699|gb|DQ37780 Squirrelpox virus 1296/99', # 1e-11
# ], result)
self.assertEqual([
'gi|887699|gb|DQ37780 Squirrelpox virus 1296/99', # 1e-11
'gi|887699|gb|DQ37780 Squirrelpox virus 55', # 1e-10
'gi|887699|gb|DQ37780 Monkeypox virus 456', # 1e-8
'gi|887699|gb|DQ37780 Mummypox virus 3000 B.C.', # 1e-7
'gi|887699|gb|DQ37780 Cowpox virus 15', # 1e-6
], result)
def testWithScoreBetterThan_EValue(self):
"""
The filter function work correctly when passed a value for
withScoreBetterThan when using e values.
"""
mockOpener = mockOpen(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n' + dumps(RECORD2) + '\n' +
dumps(RECORD3) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
reads.add(Read('id2', 'A' * 70))
reads.add(Read('id3', 'A' * 70))
readsAlignments = DiamondReadsAlignments(
reads, 'file.json', 'database.fasta',
scoreClass=LowerIsBetterScore)
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.filter(withScoreBetterThan=1e-10)
self.assertEqual(
[
'gi|887699|gb|DQ37780 Squirrelpox virus 1296/99',
],
list(result.keys()))
def testTabSeparatedSummary(self):
"""
The summary function must return the correct result.
"""
mockOpener = mockOpen(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
readsAlignments = DiamondReadsAlignments(reads, 'f.json', 'db')
titlesAlignments = TitlesAlignments(readsAlignments)
summary = titlesAlignments.tabSeparatedSummary(sortOn='title')
expected = (
'0.000297\t'
'20.000000\t'
'20.000000\t'
'1\t'
'1\t'
'37000\t'
'gi|887699|gb|DQ37780 Squirrelpox virus 1296/99'
'\n'
'0.000289\t'
'25.000000\t'
'25.000000\t'
'1\t'
'1\t'
'38000\t'
'gi|887699|gb|DQ37780 Squirrelpox virus 55')
self.assertEqual(expected, summary)
def testSummary(self):
"""
The summary function must return the correct result.
"""
mockOpener = mockOpen(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
readsAlignments = DiamondReadsAlignments(reads, 'file.json',
'database.fasta')
titlesAlignments = TitlesAlignments(readsAlignments)
self.assertEqual(
[
{
'bestScore': 20.0,
'coverage': 0.00031428571428571427,
'hspCount': 1,
'medianScore': 20.0,
'readCount': 1,
'subjectLength': 35000,
'subjectTitle': (
'gi|887699|gb|DQ37780 Monkeypox virus 456'),
},
{
'bestScore': 20.0,
'coverage': 0.00031428571428571427,
'hspCount': 1,
'medianScore': 20.0,
'readCount': 1,
'subjectLength': 35000,
'subjectTitle': (
'gi|887699|gb|DQ37780 Mummypox virus 3000 B.C.'),
},
{
'bestScore': 20.0,
'coverage': 0.0002972972972972973,
'hspCount': 1,
'medianScore': 20.0,
'readCount': 1,
'subjectLength': 37000,
'subjectTitle': (
'gi|887699|gb|DQ37780 Squirrelpox virus 1296/99'),
},
{
'bestScore': 25.0,
'coverage': 0.00028947368421052634,
'hspCount': 1,
'medianScore': 25.0,
'readCount': 1,
'subjectLength': 38000,
'subjectTitle': (
'gi|887699|gb|DQ37780 Squirrelpox virus 55'),
},
],
list(titlesAlignments.summary(sortOn='title')))
def testSummary(self):
"""
The summary function must return the correct result.
"""
mockOpener = mock_open(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
readsAlignments = DiamondReadsAlignments(reads, 'file.json')
titlesAlignments = TitlesAlignments(readsAlignments)
self.assertEqual(
[
{
'bestScore': 20.0,
'coverage': 0.00031428571428571427,
'hspCount': 1,
'medianScore': 20.0,
'readCount': 1,
'subjectLength': 35000,
'subjectTitle': (
'gi|887699|gb|DQ37780 Monkeypox virus 456'),
},
{
'bestScore': 20.0,
'coverage': 0.00031428571428571427,
'hspCount': 1,
'medianScore': 20.0,
'readCount': 1,
'subjectLength': 35000,
'subjectTitle': (
'gi|887699|gb|DQ37780 Mummypox virus 3000 B.C.'),
},
{
'bestScore': 20.0,
'coverage': 0.0002972972972972973,
'hspCount': 1,
'medianScore': 20.0,
'readCount': 1,
'subjectLength': 37000,
'subjectTitle': (
'gi|887699|gb|DQ37780 Squirrelpox virus 1296/99'),
},
{
'bestScore': 25.0,
'coverage': 0.00028947368421052634,
'hspCount': 1,
'medianScore': 25.0,
'readCount': 1,
'subjectLength': 38000,
'subjectTitle': (
'gi|887699|gb|DQ37780 Squirrelpox virus 55'),
},
],
list(titlesAlignments.summary(sortOn='title')))
def testEmpty(self):
"""
Sorting when there are no titles must return the empty list.
"""
mockOpener = mockOpen(read_data=PARAMS)
with patch.object(builtins, 'open', mockOpener):
readsAlignments = LightReadsAlignments('file.json', DB)
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.sortTitles('title')
self.assertEqual([], result)
def testEmpty(self):
"""
An instance of TitlesAlignments must have no titles if passed an
empty readsAlignments instance.
"""
mockOpener = mockOpen(read_data=(PARAMS))
with patch.object(builtins, 'open', mockOpener):
readsAlignments = LightReadsAlignments('file.json', DB)
titlesAlignments = TitlesAlignments(readsAlignments)
self.assertEqual([], list(titlesAlignments.keys()))
def testEmpty(self):
"""
Sorting when there are no titles must return the empty list.
"""
mockOpener = mock_open(read_data=dumps(PARAMS) + '\n')
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
readsAlignments = DiamondReadsAlignments(reads, 'file.json')
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.sortTitles('title')
self.assertEqual([], result)
def testEmpty(self):
"""
Sorting when there are no titles must return the empty list.
"""
mockOpener = mockOpen(read_data=dumps(PARAMS) + '\n')
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
readsAlignments = BlastReadsAlignments(reads, 'file.json')
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.sortTitles('title')
self.assertEqual([], result)
def testEmpty(self):
"""
An instance of TitlesAlignments must have no titles if passed an
empty readsAlignments instance.
"""
mockOpener = mockOpen(read_data=(dumps(PARAMS) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
readsAlignments = BlastReadsAlignments(reads, 'file.json')
titlesAlignments = TitlesAlignments(readsAlignments)
self.assertEqual([], list(titlesAlignments.keys()))
def testCoverageExcludesAll(self):
"""
The coverage function must return an titlesAlignments instance with
no titles if none of its titles has sufficient coverage.
"""
mockOpener = mockOpen(read_data=(PARAMS + RECORD0 + RECORD1 + RECORD2 +
RECORD3))
with patch.object(builtins, 'open', mockOpener):
readsAlignments = LightReadsAlignments('file.json', DB)
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.filter(minCoverage=0.1)
self.assertEqual(0, len(result))
def testEmpty(self):
"""
An instance of TitlesAlignments must have no titles if passed an
empty readsAlignments instance.
"""
mockOpener = mockOpen(read_data=(dumps(PARAMS) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
readsAlignments = DiamondReadsAlignments(reads, 'file.json',
'database.fasta')
titlesAlignments = TitlesAlignments(readsAlignments)
self.assertEqual([], list(titlesAlignments.keys()))
def testAddTitle(self):
"""
The addTitle function must add a title to the TitlesAlignments
instance.
"""
mockOpener = mockOpen(read_data=(PARAMS + RECORD0))
with patch.object(builtins, 'open', mockOpener):
readsAlignments = LightReadsAlignments('file.json', DB)
titlesAlignments = TitlesAlignments(readsAlignments)
title = 'Squirrelpox virus 23'
titleAlignments = TitleAlignments(title, 55)
self.assertTrue(title not in titlesAlignments)
titlesAlignments.addTitle(title, titleAlignments)
self.assertTrue(title in titlesAlignments)
def main(recordFilenames, fastaFilename, title, xRange, bitRange):
"""
Print reads that match in a specified X-axis and bit score range.
@param recordFilenames: A C{list} of C{str} file names contain results of a
BLAST run, in JSON format.
@param fastaFilename: The C{str} name of the FASTA file that was originally
BLASTed.
@param title: The C{str} title of the subject sequence, as output by BLAST.
@param xRange: A (start, end) list of C{int}s, giving an X-axis range or
C{None} if the entire X axis range should be printed.
@param bitRange: A (start, end) list of C{int}s, giving a bit score range
or C{None} if the entire bit score range should be printed.
"""
reads = FastaReads(fastaFilename)
blastReadsAlignments = BlastReadsAlignments(reads, recordFilenames)
filtered = blastReadsAlignments.filter(whitelist=set([title]),
negativeTitleRegex='.')
titlesAlignments = TitlesAlignments(filtered)
if title not in titlesAlignments:
print('%s: Title %r not found in BLAST output' % (sys.argv[0], title))
sys.exit(3)
for titleAlignment in titlesAlignments[title]:
for hsp in titleAlignment.hsps:
if ((xRange is None or (xRange[0] <= hsp.subjectEnd
and xRange[1] >= hsp.subjectStart))
and (bitRange is None or
(bitRange[0] <= hsp.score.score <= bitRange[1]))):
print(('query: %s, start: %d, end: %d, score: %d' %
(titleAlignment.read.id, hsp.subjectStart,
hsp.subjectEnd, hsp.score.score)))
def testTitleCollection(self):
"""
A title that occurs in the alignments of multiple reads must have
the data from those reads collected properly.
"""
mockOpener = mockOpen(read_data=(PARAMS + RECORD2 + RECORD3))
with patch.object(builtins, 'open', mockOpener):
readsAlignments = LightReadsAlignments('file.json', DB)
titlesAlignments = TitlesAlignments(readsAlignments)
title = COWPOX.id
titleAlignments = titlesAlignments[title]
self.assertEqual(2, len(titleAlignments))
self.assertEqual(title, titleAlignments.subjectTitle)
self.assertEqual(len(COWPOX.sequence),
titleAlignments.subjectLength)
self.assertEqual(READ2, titleAlignments[0].read)
self.assertEqual(READ2_COWPOX_SCORE,
titleAlignments[0].hsps[0].score.score)
self.assertEqual(READ3, titleAlignments[1].read)
self.assertEqual(READ3_COWPOX_SCORE,
titleAlignments[1].hsps[0].score.score)
def testMinMatchingReads(self):
"""
The filter function work correctly when passed a value for
minMatchingReads.
"""
mockOpener = mockOpen(read_data=(PARAMS + RECORD0 + RECORD1 + RECORD2 +
RECORD3))
with patch.object(builtins, 'open', mockOpener):
readsAlignments = LightReadsAlignments('file.json', DB)
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.filter(minMatchingReads=2)
self.assertEqual(
[
COWPOX.id,
],
list(result.keys()))
def testWithScoreBetterThan(self):
"""
The filter function work correctly when passed a value for
withScoreBetterThan.
"""
mockOpener = mockOpen(read_data=(PARAMS + RECORD0 + RECORD1 + RECORD2 +
RECORD3))
with patch.object(builtins, 'open', mockOpener):
readsAlignments = LightReadsAlignments('file.json', DB)
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.filter(withScoreBetterThan=0.9)
self.assertEqual(
[
SQUIRRELPOX.id,
],
list(result.keys()))
def testExpectedTitleDetails(self):
"""
An instance of TitleAlignments in a TitlesAlignments instance must
have the expected attributes.
"""
mockOpener = mock_open(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
read = Read('id0', 'A' * 70)
reads.add(read)
readsAlignments = DiamondReadsAlignments(reads, 'file.json')
titlesAlignments = TitlesAlignments(readsAlignments)
title = 'gi|887699|gb|DQ37780 Squirrelpox virus 1296/99'
titleAlignments = titlesAlignments[title]
self.assertEqual(title, titleAlignments.subjectTitle)
self.assertEqual(37000, titleAlignments.subjectLength)
self.assertEqual(1, len(titleAlignments))
self.assertEqual(read, titleAlignments[0].read)
self.assertEqual(HSP(20), titleAlignments[0].hsps[0])
title = 'gi|887699|gb|DQ37780 Squirrelpox virus 55'
titleAlignments = titlesAlignments[title]
self.assertEqual(title, titleAlignments.subjectTitle)
self.assertEqual(38000, titleAlignments.subjectLength)
self.assertEqual(1, len(titleAlignments))
self.assertEqual(read, titleAlignments[0].read)
self.assertEqual(HSP(25), titleAlignments[0].hsps[0])
def testTitleCollection(self):
"""
A title that occurs in the alignments of multiple reads must have
the data from both reads collected properly.
"""
mockOpener = mock_open(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD2) + '\n' +
dumps(RECORD3) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
read2 = Read('id2', 'A' * 70)
read3 = Read('id3', 'A' * 70)
reads.add(read2)
reads.add(read3)
readsAlignments = DiamondReadsAlignments(reads, 'file.json')
titlesAlignments = TitlesAlignments(readsAlignments)
title = 'gi|887699|gb|DQ37780 Cowpox virus 15'
titleAlignments = titlesAlignments[title]
self.assertEqual(title, titleAlignments.subjectTitle)
self.assertEqual(30000, titleAlignments.subjectLength)
self.assertEqual(2, len(titleAlignments))
self.assertEqual(read2, titleAlignments[0].read)
self.assertEqual(HSP(20), titleAlignments[0].hsps[0])
self.assertEqual(read3, titleAlignments[1].read)
self.assertEqual(HSP(20), titleAlignments[1].hsps[0])
def testExpectedTitleDetails(self):
"""
An instance of TitleAlignments in a TitlesAlignments instance must
have the expected attributes.
"""
mockOpener = mockOpen(read_data=(PARAMS + RECORD0))
with patch.object(builtins, 'open', mockOpener):
readsAlignments = LightReadsAlignments('file.json', DB)
titlesAlignments = TitlesAlignments(readsAlignments)
titleAlignments = titlesAlignments[SQUIRRELPOX.id]
self.assertEqual(SQUIRRELPOX.id, titleAlignments.subjectTitle)
self.assertEqual(len(SQUIRRELPOX.sequence),
titleAlignments.subjectLength)
self.assertEqual(1, len(titleAlignments))
self.assertEqual(READ0, titleAlignments[0].read)
self.assertEqual(READ0_SQUIRRELPOX_SCORE,
titleAlignments[0].hsps[0].score.score)
titleAlignments = titlesAlignments[CATPOX.id]
self.assertEqual(CATPOX.id, titleAlignments.subjectTitle)
self.assertEqual(len(CATPOX.sequence),
titleAlignments.subjectLength)
self.assertEqual(1, len(titleAlignments))
self.assertEqual(READ0, titleAlignments[0].read)
self.assertEqual(READ0_CATPOX_SCORE,
titleAlignments[0].hsps[0].score.score)
def testExpectedTitles(self):
"""
An instance of TitlesAlignments must have the expected titles.
"""
mockOpener = mockOpen(read_data=(PARAMS + RECORD0 + RECORD1 + RECORD2))
with patch.object(builtins, 'open', mockOpener):
readsAlignments = LightReadsAlignments('file.json', DB)
titlesAlignments = TitlesAlignments(readsAlignments)
self.assertEqual(
sorted([
COWPOX.id,
MONKEYPOX.id,
MUMMYPOX.id,
SQUIRRELPOX.id,
CATPOX.id,
]),
sorted(titlesAlignments.keys()))
def testTitle(self):
"""
Sorting on title must work.
"""
mockOpener = mockOpen(read_data=(PARAMS + RECORD0 + RECORD1 + RECORD2 +
RECORD3))
with patch.object(builtins, 'open', mockOpener):
readsAlignments = LightReadsAlignments('file.json', DB)
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.sortTitles('title')
self.assertEqual([
CATPOX.id,
COWPOX.id,
MONKEYPOX.id,
MUMMYPOX.id,
SQUIRRELPOX.id,
], result)
def testAddTitle(self):
"""
The addTitle function must add a title to the TitlesAlignments
instance.
"""
mockOpener = mockOpen(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
readsAlignments = BlastReadsAlignments(reads, 'file.json')
titlesAlignments = TitlesAlignments(readsAlignments)
title = 'gi|887699|gb|DQ37780 Squirrelpox virus 23'
titleAlignments = TitleAlignments(title, 55)
self.assertTrue(title not in titlesAlignments)
titlesAlignments.addTitle(title, titleAlignments)
self.assertTrue(title in titlesAlignments)
def testAddTitle(self):
"""
The addTitle function must add a title to the TitlesAlignments
instance.
"""
mockOpener = mock_open(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
readsAlignments = DiamondReadsAlignments(reads, 'file.json')
titlesAlignments = TitlesAlignments(readsAlignments)
title = 'gi|887699|gb|DQ37780 Squirrelpox virus 23'
titleAlignments = TitleAlignments(title, 55)
self.assertTrue(title not in titlesAlignments)
titlesAlignments.addTitle(title, titleAlignments)
self.assertTrue(title in titlesAlignments)
def testMaxTitlesZero(self):
"""
The filter function must return an empty result when maxTitles is zero.
"""
mockOpener = mockOpen(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n' + dumps(RECORD2) + '\n' +
dumps(RECORD3) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
reads.add(Read('id2', 'A' * 70))
reads.add(Read('id3', 'A' * 70))
readsAlignments = BlastReadsAlignments(reads, 'file.json')
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.filter(maxTitles=0, sortOn='maxScore')
self.assertEqual(0, len(result))
def testCoverageExcludesAll(self):
"""
The coverage function must return an titlesAlignments instance with
no titles if none of its titles has sufficient coverage.
"""
mockOpener = mockOpen(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n' + dumps(RECORD2) + '\n' +
dumps(RECORD3) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
reads.add(Read('id2', 'A' * 70))
reads.add(Read('id3', 'A' * 70))
readsAlignments = BlastReadsAlignments(reads, 'file.json')
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.filter(minCoverage=0.1)
self.assertEqual(0, len(result))
def testHsps(self):
"""
The hsps function must yield all the hsps for all titles in a
TitlesAlignments instance.
"""
mockOpener = mockOpen(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n' + dumps(RECORD2) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
reads.add(Read('id2', 'A' * 70))
readsAlignments = BlastReadsAlignments(reads, 'file.json')
titlesAlignments = TitlesAlignments(readsAlignments)
result = list(titlesAlignments.hsps())
self.assertEqual(
sorted([HSP(20), HSP(25), HSP(20), HSP(20), HSP(20)]),
sorted(result))
def testReadSetFilterStrict(self):
"""
The filter function work correctly when passed a 1.0 value for
minNewReads.
"""
mockOpener = mockOpen(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n' + dumps(RECORD2) + '\n' +
dumps(RECORD3) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
reads.add(Read('id2', 'A' * 70))
reads.add(Read('id3', 'A' * 70))
readsAlignments = DiamondReadsAlignments(reads, 'file.json',
'database.fasta')
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.filter(minNewReads=1.0)
# Either 'gi|887699|gb|DQ37780 Mummypox virus 3000 B.C.'
# invalidates 'gi|887699|gb|DQ37780 Monkeypox virus 456' or
# vice-versa. It depends on Python's dict walking order. Check
# for both, making sure just one of them is true.
mummypox = 'gi|887699|gb|DQ37780 Mummypox virus 3000 B.C.'
monkeypox = 'gi|887699|gb|DQ37780 Monkeypox virus 456'
assertionCount = 0
if mummypox in result:
self.assertTrue(monkeypox in
result.readSetFilter.invalidates(mummypox))
assertionCount += 1
if monkeypox in result:
self.assertTrue(mummypox in
result.readSetFilter.invalidates(monkeypox))
assertionCount += 1
self.assertEqual(1, assertionCount)
def testExpectedTitles(self):
"""
An instance of TitlesAlignments must have the expected titles.
"""
mockOpener = mockOpen(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n' + dumps(RECORD2) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
reads.add(Read('id2', 'A' * 70))
readsAlignments = BlastReadsAlignments(reads, 'file.json')
titlesAlignments = TitlesAlignments(readsAlignments)
self.assertEqual(
[
'gi|887699|gb|DQ37780 Cowpox virus 15',
'gi|887699|gb|DQ37780 Monkeypox virus 456',
'gi|887699|gb|DQ37780 Mummypox virus 3000 B.C.',
'gi|887699|gb|DQ37780 Squirrelpox virus 1296/99',
'gi|887699|gb|DQ37780 Squirrelpox virus 55',
],
sorted(titlesAlignments.keys()))
def testWithScoreBetterThan_Bits(self):
"""
The filter function work correctly when passed a value for
withScoreBetterThan when using bit scores.
"""
mockOpener = mockOpen(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n' + dumps(RECORD2) + '\n' +
dumps(RECORD3) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
reads.add(Read('id2', 'A' * 70))
reads.add(Read('id3', 'A' * 70))
readsAlignments = BlastReadsAlignments(reads, 'file.json')
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.filter(withScoreBetterThan=24)
self.assertEqual(
[
'gi|887699|gb|DQ37780 Squirrelpox virus 55',
],
list(result.keys()))
def testMinMedianScore_Bits(self):
"""
The filter function work correctly when passed a value for
minMedianScore when using bit scores.
"""
mockOpener = mockOpen(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n' + dumps(RECORD2) + '\n' +
dumps(RECORD3) + '\n'))
with patch('__builtin__.open', mockOpener, create=True):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
reads.add(Read('id2', 'A' * 70))
reads.add(Read('id3', 'A' * 70))
readsAlignments = BlastReadsAlignments(reads, 'file.json')
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.filter(minMedianScore=22)
self.assertEqual(
[
'gi|887699|gb|DQ37780 Squirrelpox virus 55',
],
result.keys())
def testMaxTitlesOne(self):
"""
The filter function must return just the best title when maxTitles
is one.
"""
mockOpener = mockOpen(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n' + dumps(RECORD2) + '\n' +
dumps(RECORD3) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
reads.add(Read('id2', 'A' * 70))
reads.add(Read('id3', 'A' * 70))
readsAlignments = BlastReadsAlignments(reads, 'file.json')
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.filter(maxTitles=1, sortOn='maxScore')
self.assertEqual(
[
'gi|887699|gb|DQ37780 Squirrelpox virus 55',
],
sorted(result.keys()))
def testMinMatchingReads(self):
"""
The filter function work correctly when passed a value for
minMatchingReads.
"""
mockOpener = mockOpen(read_data=(
dumps(PARAMS) + '\n' + dumps(RECORD0) + '\n' +
dumps(RECORD1) + '\n' + dumps(RECORD2) + '\n' +
dumps(RECORD3) + '\n'))
with patch.object(builtins, 'open', mockOpener):
reads = Reads()
reads.add(Read('id0', 'A' * 70))
reads.add(Read('id1', 'A' * 70))
reads.add(Read('id2', 'A' * 70))
reads.add(Read('id3', 'A' * 70))
readsAlignments = BlastReadsAlignments(reads, 'file.json')
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.filter(minMatchingReads=2)
self.assertEqual(
[
'gi|887699|gb|DQ37780 Cowpox virus 15',
],
list(result.keys()))
maxSequenceLen=args.maxSequenceLen,
minStart=args.minStart,
maxStop=args.maxStop,
oneAlignmentPerRead=args.oneAlignmentPerRead,
maxHspsPerHit=args.maxHspsPerHit,
scoreCutoff=args.scoreCutoff,
whitelist=set(args.whitelist) if args.whitelist else None,
blacklist=set(args.blacklist) if args.blacklist else None,
titleRegex=args.titleRegex,
negativeTitleRegex=args.negativeTitleRegex,
truncateTitlesAfter=args.truncateTitlesAfter,
taxonomy=args.taxonomy)
titlesAlignments = TitlesAlignments(readsAlignments).filter(
minMatchingReads=args.minMatchingReads,
minMedianScore=args.minMedianScore,
withScoreBetterThan=args.withScoreBetterThan,
minNewReads=args.minNewReads)
nTitles = len(titlesAlignments)
print('Found %d interesting title%s.' % (nTitles,
'' if nTitles == 1 else 's'))
if args.earlyExit:
print('Matched titles (sorted by best score, descending):')
print('\n'.join(titlesAlignments.sortTitles('maxScore')))
sys.exit(0)
alignmentPanel(titlesAlignments, sortOn=args.sortOn, interactive=True,
outputDir=args.outputDir,
idList=parseColors(args.color) if args.color else None,
reads, jsonFiles, args.diamondDatabaseFastaFilename)
readsAlignments.filter(
minSequenceLen=args.minSequenceLen,
maxSequenceLen=args.maxSequenceLen,
minStart=args.minStart, maxStop=args.maxStop,
oneAlignmentPerRead=args.oneAlignmentPerRead,
maxHspsPerHit=args.maxHspsPerHit,
scoreCutoff=args.scoreCutoff,
whitelist=whitelist, blacklist=blacklist,
titleRegex=args.titleRegex, negativeTitleRegex=args.negativeTitleRegex,
truncateTitlesAfter=args.truncateTitlesAfter, taxonomy=args.taxonomy)
titlesAlignments = TitlesAlignments(readsAlignments).filter(
minMatchingReads=args.minMatchingReads,
minMedianScore=args.minMedianScore,
withScoreBetterThan=args.withScoreBetterThan,
minNewReads=args.minNewReads, maxTitles=args.maxTitles,
sortOn=args.sortOn, minCoverage=args.minCoverage)
nTitles = len(titlesAlignments)
print('Found %d interesting title%s.' %
(nTitles, '' if nTitles == 1 else 's'), file=sys.stderr)
if nTitles:
print(titlesAlignments.tabSeparatedSummary(sortOn=args.sortOn))
if args.earlyExit:
sys.exit(0)
if nTitles == 0:
print('No alignment panel generated due to no matching titles.',
scoreCutoff=args.scoreCutoff,
whitelist=args.whitelist,
blacklist=args.blacklist,
titleRegex=args.titleRegex,
negativeTitleRegex=args.negativeTitleRegex,
truncateTitlesAfter=args.truncateTitlesAfter,
taxonomy=args.taxonomy)
reads = Reads()
if (args.minMatchingReads is None and args.minMedianScore is None and
args.withScoreBetterThan is None and args.minNewReads is None):
# No need to collect into titles, just get the read ids from
# the matching alignments.
for readAlignment in readsAlignments:
reads.add(readAlignment.read)
else:
# We need to collect alignments into titles.
titlesAlignments = TitlesAlignments(readsAlignments).filter(
minMatchingReads=args.minMatchingReads,
minMedianScore=args.minMedianScore,
withScoreBetterThan=args.withScoreBetterThan,
minNewReads=args.minNewReads)
for titleAlignments in titlesAlignments.values():
for alignment in titleAlignments.alignments:
reads.add(alignment.read)
reads.save(sys.stdout)
print('Found %d matching reads.' % len(reads), file=sys.stderr)
scoreCutoff=args.scoreCutoff,
whitelist=args.whitelist,
blacklist=args.blacklist,
titleRegex=args.titleRegex,
negativeTitleRegex=args.negativeTitleRegex,
truncateTitlesAfter=args.truncateTitlesAfter,
taxonomy=args.taxonomy)
reads = Reads()
if (args.minMatchingReads is None and args.minMedianScore is None and
args.withScoreBetterThan is None and args.minNewReads is None):
# No need to collect into titles, just get the read ids from
# the matching alignments.
for readAlignment in readsAlignments:
reads.add(readAlignment.read)
else:
# We need to collect alignments into titles.
titlesAlignments = TitlesAlignments(readsAlignments).filter(
minMatchingReads=args.minMatchingReads,
minMedianScore=args.minMedianScore,
withScoreBetterThan=args.withScoreBetterThan,
minNewReads=args.minNewReads)
for titleAlignments in titlesAlignments.itervalues():
for alignment in titleAlignments.alignments:
reads.add(alignment.read)
reads.save(sys.stdout)
print >>sys.stderr, 'Found %d matching reads.' % len(reads)
