def testSubjectsCompareEqualEvenWithDifferingHashCounts(self):
"""
Two otherwise identical Subject instances must compare equal even if
their hash counts differ.
"""
read = AARead('id', 'AA')
self.assertEqual(Subject(read, 1), Subject(read, 2))
def testSubjectsOfIncompatibleReadTypesCompareUnequal(self):
"""
Two Subject instances that have reads that cannot be compared must
compare unequal.
"""
read1 = AARead('id', 'AA')
read2 = SSAARead('id', 'AA', 'HH')
self.assertNotEqual(Subject(read1), Subject(read2))
def testAddTwoSubjectsWithIndex(self):
"""
Adding two subjects to a SubjectStore and giving their indices
must have the expected result.
"""
ss = SubjectStore()
subject1 = Subject(AARead('id1', 'AA'))
subject2 = Subject(AARead('id2', 'AA'))
ss.add(subject1, '1')
ss.add(subject2, '2')
self.assertEqual(2, len(ss))
def addSubject(self, subject):
"""
Ask a backend to add a subject.
@param subject: A C{dark.read.AARead} instance. The subject sequence
is passed as a read instance even though in many cases it will not
be an actual read from a sequencing run.
@raises BackendException: if no backends are available.
@return: A tuple of 1) a C{bool} to indicate whether the subject was
already in the database, and 2) the C{str} subject index.
"""
if not self._backends:
raise BackendException('Database has no backends.')
# We could check to see if self._disconnectedBackends is non-empty
# and issue a warning if so. But there's no reason we can't add a
# subject to one of the connected backends.
subject = Subject(subject.id, subject.sequence, 0, subject.quality)
preExisting, subjectIndex = self._subjectStore.add(subject)
if preExisting:
return True, subjectIndex
# Pick a backend to add the subject to. Do this by choosing a
# session id at random.
sessionId = choice([self._sessionIdToName.keys()])
preExisting, BackendSubjectIndex, hashCount = (yield from
self._component.call(
'addSubject-%s' %
sessionId,
subject.id,
subject.sequence,
subject.quality))
# Sanity check that the backend had not seen the subject before and
# that it used the subjectIndex we gave it.
assert (preExisting is False)
assert (subjectIndex == BackendSubjectIndex)
# Store the hash count returned by the backend so that our subject
# store is accurate.
subject.hashCount = hashCount
# Update our own checksum and that of the specific backend to which
# the subject was added.
name = self._sessionIdToName[sessionId]
update = (subject.id, subject.sequence)
self._checksum.update(update)
self._backends[name]['checksum'].update(update)
self._backends[name]['subjectCount'] += 1
return False, subjectIndex
def testGetSubjects(self):
"""
getSubjects must return the expected result.
"""
ss = SubjectStore()
subject1 = Subject(AARead('id1', 'AA'))
subject2 = Subject(AARead('id2', 'CC'))
ss.add(subject1, '1')
ss.add(subject2, '2')
result = set(ss.getSubjects())
self.assertEqual(set([subject1, subject2]), result)
def testAddSubjectWithIndex(self):
"""
Adding one subject to a SubjectStore and giving a subject index
must return the expected result.
"""
subject = Subject(AARead('id', 'AA'))
preExisting, subjectIndex = SubjectStore().add(subject, '3')
self.assertFalse(preExisting)
self.assertEqual('3', subjectIndex)
def testLengthIsOneAfterAddSubject(self):
"""
Adding a subject to a new SubjectStore results in SubjectStore with
length one.
"""
ss = SubjectStore()
subject = Subject(AARead('id', 'AA'))
ss.add(subject)
self.assertEqual(1, len(ss))
def testLengthIsOneAfterAddSameSubjectTwice(self):
"""
Adding a subject to a new SubjectStore twice (with an identical index)
results in SubjectStore with length one.
"""
ss = SubjectStore()
subject = Subject(AARead('id', 'AA'))
ss.add(subject)
ss.add(subject)
self.assertEqual(1, len(ss))
def testGetIndexBySubjectKeyError(self):
"""
If a non-existent subject is passed to getIndexBySubject, a KeyError
must be raised.
"""
dbParams = DatabaseParameters(landmarks=[AlphaHelix], trigPoints=[])
db = Database(dbParams)
error = "^'id'$"
six.assertRaisesRegex(self, KeyError, error, db.getIndexBySubject,
Subject(AARead('id', 'FF')))
def testGetSubjectBySubject(self):
"""
If a subject is added, getIndexBySubject must be able to return it
given an identical subject to look up.
"""
dbParams = DatabaseParameters(landmarks=[AlphaHelix], trigPoints=[])
db = Database(dbParams)
_, index, _ = db.addSubject(AARead('id', 'FRRRFRRRF'))
self.assertEqual(
index, db.getIndexBySubject(Subject(AARead('id', 'FRRRFRRRF'))))
def testGetSubjectByIndex(self):
"""
If a subject is added, getSubjectByIndex must be able to return it
given its string index.
"""
dbParams = DatabaseParameters(landmarks=[AlphaHelix], trigPoints=[])
db = Database(dbParams)
subject = AARead('id', 'FRRRFRRRF')
_, index, _ = db.addSubject(subject)
self.assertEqual(Subject(AARead('id', 'FRRRFRRRF')),
db.getSubjectByIndex(index))
def testOneReadTwoLandmarksGetSubjects(self):
"""
If one subject with two landmarks (and hence one hash) is added, an
entry is appended to the database subject info.
"""
dbParams = DatabaseParameters(landmarks=[AlphaHelix], trigPoints=[])
db = Database(dbParams)
db.addSubject(AARead('id', 'FRRRFRRRFAFRRRFRRRF'))
subject = list(db.getSubjects())[0]
read = AARead('id', 'FRRRFRRRFAFRRRFRRRF')
self.assertEqual(Subject(read, 1), subject)
self.assertEqual(1, subject.hashCount)
def testAddSubjectTwiceWithSameIndex(self):
"""
Adding a subject to a SubjectStore and giving a subject index
and then re-adding it with the same index must return the expected
result.
"""
subject = Subject(AARead('id', 'AA'))
ss = SubjectStore()
ss.add(subject, '3')
preExisting, subjectIndex = ss.add(subject, '3')
self.assertTrue(preExisting)
self.assertEqual('3', subjectIndex)
def testAddSubjectTwiceWithDifferentIndex(self):
"""
Adding a subject to a SubjectStore and giving a subject index
and then re-adding it with a different index must raise a
SubjectStoreException.
"""
subject = Subject(AARead('id', 'AA'))
ss = SubjectStore()
ss.add(subject, '3')
error = ("^Already stored index \(3\) for subject 'id' does not match "
"subsequently passed index \(4\) for the subject\.$")
six.assertRaisesRegex(self, SubjectStoreException, error, ss.add,
subject, '4')
def testOneReadOneLandmarkGetSubjects(self):
"""
If one subject with just one landmark (and hence no hashes) is added,
an entry is appended to the database subject info.
"""
dbParams = DatabaseParameters(landmarks=[AlphaHelix], trigPoints=[])
db = Database(dbParams)
db.addSubject(AARead('id', 'FRRRFRRRF'))
subjects = list(db.getSubjects())
self.assertEqual(1, len(subjects))
subject = subjects[0]
self.assertEqual(Subject(AARead('id', 'FRRRFRRRF')), subject)
self.assertEqual(0, subject.hashCount)
def testSaveRestoreNonEmpty(self):
"""
After a save/restore of a non-empty subject store, the subject store
must be as expected.
"""
ss = SubjectStore()
subject = Subject(AARead('id', 'AA'), 6)
ss.add(subject, '3')
fp = StringIO()
ss.save(fp)
fp.seek(0)
ss = SubjectStore.restore(fp)
self.assertEqual(1, len(ss))
self.assertEqual('3', ss.getIndexBySubject(subject))
result = ss.getSubjectByIndex('3')
self.assertEqual(subject, result)
self.assertEqual(6, result.hashCount)
def testHashCountIsStored(self):
"""
A Subject must store the hashcount it is passed.
"""
read = AARead('id', 'AA')
self.assertEqual(6, Subject(read, 6).hashCount)
def testHashCountDefaultsToZero(self):
"""
A Subject must store a hashcount of zero if the hashcount isn't given.
"""
read = AARead('id', 'AA')
self.assertEqual(0, Subject(read).hashCount)
def testLen(self):
"""
A Subject must return the correct (sequence) length via len().
"""
self.assertEqual(2, len(Subject(AARead('id', 'AA'))))
def testIdenticalSubjectsCompareEqual(self):
"""
Two identical Subject instances must compare equal.
"""
read = AARead('id', 'AA')
self.assertEqual(Subject(read), Subject(read))
