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 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 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 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 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 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 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 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 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 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 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 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 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 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 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 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 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 testMaxTitlesZero(self):
"""
The filter function must return an empty result when maxTitles is zero.
"""
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(maxTitles=0, sortOn='maxScore')
self.assertEqual(0, len(result))
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 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=(PARAMS + RECORD0 + RECORD1 + RECORD2))
with patch.object(builtins, 'open', mockOpener):
readsAlignments = LightReadsAlignments('file.json', DB)
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.filter()
self.assertEqual(
sorted([
COWPOX.id,
MONKEYPOX.id,
MUMMYPOX.id,
SQUIRRELPOX.id,
CATPOX.id,
]),
sorted(iter(result)))
def testCoverageExcludesAll(self):
"""
The coverage function must return an titlesAlignments instance with
no titles if none of 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=0.1)
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 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=(PARAMS + RECORD0 + RECORD1 + RECORD2 +
RECORD3))
with patch.object(builtins, 'open', mockOpener):
readsAlignments = LightReadsAlignments('file.json', DB)
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.filter(minNewReads=0.0)
self.assertEqual(sorted(
[
CATPOX.id,
COWPOX.id,
MONKEYPOX.id,
MUMMYPOX.id,
SQUIRRELPOX.id,
]),
sorted(iter(result)))
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=(PARAMS + RECORD0 + RECORD1 + RECORD2 +
RECORD3))
with patch.object(builtins, 'open', mockOpener):
readsAlignments = LightReadsAlignments('file.json', DB)
titlesAlignments = TitlesAlignments(readsAlignments)
result = titlesAlignments.filter(minCoverage=0.0)
self.assertEqual(
[
COWPOX.id,
MONKEYPOX.id,
MUMMYPOX.id,
SQUIRRELPOX.id,
CATPOX.id,
],
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()))
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 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 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=(PARAMS + RECORD0 + RECORD1 + RECORD2 +
RECORD3))
with patch.object(builtins, 'open', mockOpener):
readsAlignments = LightReadsAlignments('file.json', DB)
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(
[
COWPOX.id,
MONKEYPOX.id,
MUMMYPOX.id,
],
sorted(result.keys()))
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 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 = BlastReadsAlignments(reads, 'file.json')
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 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 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 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 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()))
