def test_alignment_reference(self):
rfn = data.getXml(9)
rs1 = ReferenceSet(data.getXml(9))
fasta_res = rs1.externalResources[0]
fasta_file = urlparse(fasta_res.resourceId).path
ds1 = AlignmentSet(data.getXml(8),
referenceFastaFname=rs1)
aln_ref = None
for aln in ds1:
aln_ref = aln.reference()
break
self.assertTrue(aln_ref is not None)
self.assertEqual(ds1.externalResources[0].reference, fasta_file)
self.assertEqual(ds1.resourceReaders()[0].referenceFasta.filename,
fasta_file)
ds1 = AlignmentSet(data.getXml(8),
referenceFastaFname=fasta_file)
aln_ref = None
for aln in ds1:
aln_ref = aln.reference()
break
self.assertTrue(aln_ref is not None)
self.assertEqual(ds1.externalResources[0].reference, fasta_file)
self.assertEqual(ds1.resourceReaders()[0].referenceFasta.filename,
fasta_file)
ds1 = AlignmentSet(data.getXml(8))
ds1.addReference(fasta_file)
aln_ref = None
for aln in ds1:
aln_ref = aln.reference()
break
self.assertTrue(aln_ref is not None)
self.assertEqual(ds1.externalResources[0].reference, fasta_file)
self.assertEqual(ds1.resourceReaders()[0].referenceFasta.filename,
fasta_file)
fofn_out = tempfile.NamedTemporaryFile(suffix=".fofn").name
log.debug(fofn_out)
with open(fofn_out, 'w') as f:
f.write(data.getXml(8))
f.write('\n')
f.write(data.getXml(11))
f.write('\n')
ds1 = AlignmentSet(fofn_out,
referenceFastaFname=fasta_file)
aln_ref = None
for aln in ds1:
aln_ref = aln.reference()
break
self.assertTrue(aln_ref is not None)
self.assertEqual(ds1.externalResources[0].reference, fasta_file)
self.assertEqual(ds1.resourceReaders()[0].referenceFasta.filename,
fasta_file)
def test_alignment_reference(self):
rfn = data.getXml(9)
rs1 = ReferenceSet(data.getXml(9))
fasta_res = rs1.externalResources[0]
fasta_file = urlparse(fasta_res.resourceId).path
ds1 = AlignmentSet(data.getXml(8), referenceFastaFname=rs1)
aln_ref = None
for aln in ds1:
aln_ref = aln.reference()
break
self.assertTrue(aln_ref is not None)
self.assertEqual(ds1.externalResources[0].reference, fasta_file)
self.assertEqual(ds1.resourceReaders()[0].referenceFasta.filename,
fasta_file)
ds1 = AlignmentSet(data.getXml(8), referenceFastaFname=fasta_file)
aln_ref = None
for aln in ds1:
aln_ref = aln.reference()
break
self.assertTrue(aln_ref is not None)
self.assertEqual(ds1.externalResources[0].reference, fasta_file)
self.assertEqual(ds1.resourceReaders()[0].referenceFasta.filename,
fasta_file)
ds1 = AlignmentSet(data.getXml(8))
ds1.addReference(fasta_file)
aln_ref = None
for aln in ds1:
aln_ref = aln.reference()
break
self.assertTrue(aln_ref is not None)
self.assertEqual(ds1.externalResources[0].reference, fasta_file)
self.assertEqual(ds1.resourceReaders()[0].referenceFasta.filename,
fasta_file)
fofn_out = tempfile.NamedTemporaryFile(suffix=".fofn").name
log.debug(fofn_out)
with open(fofn_out, 'w') as f:
f.write(data.getXml(8))
f.write('\n')
f.write(data.getXml(11))
f.write('\n')
ds1 = AlignmentSet(fofn_out, referenceFastaFname=fasta_file)
aln_ref = None
for aln in ds1:
aln_ref = aln.reference()
break
self.assertTrue(aln_ref is not None)
self.assertEqual(ds1.externalResources[0].reference, fasta_file)
self.assertEqual(ds1.resourceReaders()[0].referenceFasta.filename,
fasta_file)
def test_loading_reference(self):
log.info('Opening Reference')
r = ReferenceSet(data.getRef()).toExternalFiles()[0]
log.info('Done Opening Reference')
log.info('Opening AlignmentSet')
d = AlignmentSet(data.getBam(), referenceFastaFname=r)
log.info('Done Opening AlignmentSet')
bfile = openIndexedAlignmentFile(data.getBam(), referenceFastaFname=r)
self.assertTrue(bfile.isReferenceLoaded)
for res in d.resourceReaders():
self.assertTrue(res.isReferenceLoaded)
aln = AlignmentSet(data.getBam())
aln.addReference(r)
for res in aln.resourceReaders():
self.assertTrue(res.isReferenceLoaded)
def test_readGroupTable(self):
aln = AlignmentSet(data.getBam(0), data.getBam(1), data.getBam(2))
readers = aln.resourceReaders()
self.assertEqual(len(readers[0].readGroupTable), 1)
self.assertEqual(len(readers[1].readGroupTable), 1)
self.assertEqual(len(readers[2].readGroupTable), 1)
self.assertEqual(len(aln.readGroupTable), 3)
def test_loading_reference(self):
log.info('Opening Reference')
r = ReferenceSet(data.getRef()).toExternalFiles()[0]
log.info('Done Opening Reference')
log.info('Opening AlignmentSet')
d = AlignmentSet(data.getBam(), referenceFastaFname=r)
log.info('Done Opening AlignmentSet')
bfile = openIndexedAlignmentFile(data.getBam(),
referenceFastaFname=r)
self.assertTrue(bfile.isReferenceLoaded)
for res in d.resourceReaders():
self.assertTrue(res.isReferenceLoaded)
aln = AlignmentSet(data.getBam())
aln.addReference(r)
for res in aln.resourceReaders():
self.assertTrue(res.isReferenceLoaded)
def test_readGroupTable(self):
aln = AlignmentSet(data.getBam(0), data.getBam(1), data.getBam(2))
readers = aln.resourceReaders()
self.assertEqual(len(readers[0].readGroupTable), 1)
self.assertEqual(len(readers[1].readGroupTable), 1)
self.assertEqual(len(readers[2].readGroupTable), 1)
self.assertEqual(len(aln.readGroupTable), 3)
def test_referenceInfo(self):
aln = AlignmentSet(data.getBam(0))
readers = aln.resourceReaders()
self.assertEqual(len(readers[0].referenceInfoTable), 59)
self.assertEqual(
str(readers[0].referenceInfo('E.faecalis.1')),
"(27, 27, 'E.faecalis.1', 'E.faecalis.1', 1482, "
"'a1a59c267ac1341e5a12bce7a7d37bcb', 0L, 0L)")
def test_referenceInfo(self):
aln = AlignmentSet(data.getBam(0))
readers = aln.resourceReaders()
self.assertEqual(len(readers[0].referenceInfoTable), 59)
self.assertEqual(
str(readers[0].referenceInfo('E.faecalis.1')),
"(27, 27, 'E.faecalis.1', 'E.faecalis.1', 1482, "
"'a1a59c267ac1341e5a12bce7a7d37bcb', 0L, 0L)")
def test_referenceInfoTable(self):
aln = AlignmentSet(data.getBam(0), data.getBam(1), data.getBam(2))
readers = aln.resourceReaders()
self.assertEqual(len(readers[0].referenceInfoTable), 1)
self.assertEqual(len(readers[1].referenceInfoTable), 59)
self.assertEqual(len(readers[2].referenceInfoTable), 1)
self.assertEqual(readers[0].referenceInfoTable.Name,
readers[2].referenceInfoTable.Name)
self.assertEqual(len(aln.referenceInfoTable), 60)
def test_referenceInfoTable(self):
aln = AlignmentSet(data.getBam(0), data.getBam(1), data.getBam(2))
readers = aln.resourceReaders()
self.assertEqual(len(readers[0].referenceInfoTable), 1)
self.assertEqual(len(readers[1].referenceInfoTable), 59)
self.assertEqual(len(readers[2].referenceInfoTable), 1)
self.assertEqual(readers[0].referenceInfoTable.Name,
readers[2].referenceInfoTable.Name)
self.assertEqual(len(aln.referenceInfoTable), 60)
def test_updateCounts(self):
log.info("Testing updateCounts without filters")
aln = AlignmentSet(data.getBam(0))
readers = aln.resourceReaders()
expLen = 0
for reader in readers:
for record in reader:
expLen += record.readLength
self.assertEqual(
record.aStart, record.bam.pbi[record.rowNumber]['aStart'])
self.assertEqual(
record.aEnd, record.bam.pbi[record.rowNumber]['aEnd'])
expNum = 0
for reader in readers:
expNum += len(reader)
accLen = aln.metadata.totalLength
accNum = aln.metadata.numRecords
self.assertEqual(expLen, accLen)
self.assertEqual(expNum, accNum)
log.info("Testing whether filters are respected")
aln.filters.addRequirement(rname=[('=', 'E.faecalis.1')])
aln.updateCounts()
accLen = aln.metadata.totalLength
accNum = aln.metadata.numRecords
def count(gen):
count = 0
for _ in gen:
count += 1
return count
expLen = 0
for reader in readers:
for record in reader:
expLen += record.readLength
bfile = openIndexedAlignmentFile(data.getBam(0))
rWin = (bfile.referenceInfo('E.faecalis.1').ID,
0,
bfile.referenceInfo('E.faecalis.1').Length)
reads = bfile.readsInRange(*rWin)
expNum = count(reads)
expLen = 0
reads = bfile.readsInRange(*rWin)
for read in reads:
expLen += read.readLength
self.assertEqual(expLen, accLen)
self.assertEqual(expNum, accNum)
def test_updateCounts(self):
log.info("Testing updateCounts without filters")
aln = AlignmentSet(data.getBam(0))
readers = aln.resourceReaders()
expLen = 0
for reader in readers:
for record in reader:
expLen += record.readLength
self.assertEqual(record.aStart,
record.bam.pbi[record.rowNumber]['aStart'])
self.assertEqual(record.aEnd,
record.bam.pbi[record.rowNumber]['aEnd'])
expNum = 0
for reader in readers:
expNum += len(reader)
accLen = aln.metadata.totalLength
accNum = aln.metadata.numRecords
self.assertEqual(expLen, accLen)
self.assertEqual(expNum, accNum)
log.info("Testing whether filters are respected")
aln.filters.addRequirement(rname=[('=', 'E.faecalis.1')])
aln.updateCounts()
accLen = aln.metadata.totalLength
accNum = aln.metadata.numRecords
def count(gen):
count = 0
for _ in gen:
count += 1
return count
expLen = 0
for reader in readers:
for record in reader:
expLen += record.readLength
bfile = openIndexedAlignmentFile(data.getBam(0))
rWin = (bfile.referenceInfo('E.faecalis.1').ID, 0,
bfile.referenceInfo('E.faecalis.1').Length)
reads = bfile.readsInRange(*rWin)
expNum = count(reads)
expLen = 0
reads = bfile.readsInRange(*rWin)
for read in reads:
expLen += read.readLength
self.assertEqual(expLen, accLen)
self.assertEqual(expNum, accNum)
def test_referenceInfoTableMerging(self):
log.info("Testing refIds, etc. after merging")
ds = DataSet(data.getXml(17))
also_lambda = ds.toExternalFiles()[0]
aln = AlignmentSet(data.getBam(0), data.getBam(0), also_lambda)
readers = aln.resourceReaders()
ids = sorted([i for _, i in aln.refInfo('ID')])
self.assertEqual(range(len(ids)), ids)
accNames = aln.refNames
expNames = reduce(
np.append,
[reader.referenceInfoTable['Name'] for reader in readers])
expNames = np.unique(expNames)
self.assertEqual(sorted(expNames), sorted(accNames))
accNames = aln.fullRefNames
expNames = reduce(
np.append,
[reader.referenceInfoTable['FullName'] for reader in readers])
expNames = np.unique(expNames)
self.assertEqual(sorted(expNames), sorted(accNames))
def test_referenceInfoTableMerging(self):
log.info("Testing refIds, etc. after merging")
ds = DataSet(data.getXml(17))
also_lambda = ds.toExternalFiles()[0]
aln = AlignmentSet(data.getBam(0), data.getBam(0), also_lambda)
readers = aln.resourceReaders()
ids = sorted([i for _, i in aln.refInfo('ID')])
self.assertEqual(range(len(ids)), ids)
accNames = aln.refNames
expNames = reduce(np.append,
[reader.referenceInfoTable['Name']
for reader in readers])
expNames = np.unique(expNames)
self.assertEqual(sorted(expNames), sorted(accNames))
accNames = aln.fullRefNames
expNames = reduce(np.append,
[reader.referenceInfoTable['FullName']
for reader in readers])
expNames = np.unique(expNames)
self.assertEqual(sorted(expNames), sorted(accNames))
def _get_readers(self):
ds_reader = AlignmentSet(self.aln_path, strict=True,
reference=self.ref_path)
return ds_reader, ds_reader.resourceReaders()
def _get_pbi(self):
ds = AlignmentSet(get_file("aligned-xml"))
return ds.resourceReaders()[0].pbi
def _get_readers(self):
ds_reader = AlignmentSet(self.aln_path,
strict=True,
reference=self.ref_path)
return ds_reader, ds_reader.resourceReaders()
