def __init__(self, gaReadGroupSet, samFileName):
filename = os.path.split(samFileName)[1]
localId = os.path.splitext(filename)[0]
self.gaReadGroup = reads.AbstractReadGroup(gaReadGroupSet, localId)
self.id = self.gaReadGroup.getId()
self.samFile = pysam.AlignmentFile(samFileName)
self.reads = []
self.refIds = collections.defaultdict(list)
for read in self.samFile:
if read.reference_id != -1:
# mapped read
refId = self.samFile.getrname(read.reference_id)
self.refIds[refId].append(read)
self.reads.append(read)
def testGetReadGroup(self):
path = "/readgroups"
for dataset in self.backend.getDatasets():
for readGroupSet in dataset.getReadGroupSets():
for readGroup in readGroupSet.getReadGroups():
responseObject = self.sendGetObject(
path, readGroup.getId(), protocol.ReadGroup)
self.verifyReadGroupsEqual(responseObject, readGroup)
for badId in self.getBadIds():
readGroup = reads.AbstractReadGroup(readGroupSet, badId)
self.verifyGetMethodFails(path, readGroup.getId())
for badId in self.getBadIds():
readGroupSet = reads.AbstractReadGroupSet(dataset, badId)
self.verifyGetMethodFails(path, readGroupSet.getId())
for badId in self.getBadIds():
self.verifyGetMethodFails(path, badId)
def __init__(self, gaReadGroupSet, samFile, readGroupName):
self.gaReadGroup = reads.AbstractReadGroup(gaReadGroupSet,
readGroupName)
self.id = self.gaReadGroup.getId()
self.samFile = samFile
self.mappedReads = collections.defaultdict(list)
for read in self.samFile:
tags = dict(read.tags)
if 'RG' not in tags or tags['RG'] != readGroupName:
continue
if read.reference_id != -1:
# mapped read
referenceName = self.samFile.getrname(read.reference_id)
self.mappedReads[referenceName].append(read)
self.numAlignedReads = -1
self.numUnalignedReads = -1
self.programs = []
if 'PG' in self.samFile.header:
self.programs = self.samFile.header['PG']
self.sampleId = None
self.description = None
self.predictedInsertSize = None
self.instrumentModel = None
self.sequencingCenter = None
self.experimentDescription = None
self.library = None
self.platformUnit = None
self.runTime = None
if 'RG' in self.samFile.header:
readGroupHeader = [
rgHeader for rgHeader in self.samFile.header['RG']
if rgHeader['ID'] == readGroupName
][0]
self.sampleId = readGroupHeader.get('SM', None)
self.description = readGroupHeader.get('DS', None)
if 'PI' in readGroupHeader:
self.predictedInsertSize = int(readGroupHeader['PI'])
self.instrumentModel = readGroupHeader.get('PL', None)
self.sequencingCenter = readGroupHeader.get('CN', None)
self.experimentDescription = readGroupHeader.get('DS', None)
self.library = readGroupHeader.get('LB', None)
self.platformUnit = readGroupHeader.get('PU', None)
self.runTime = readGroupHeader.get('DT', None)
def testNonexistantReadGroup(self):
# a request for a readGroup that doesn't exist should throw an error
readGroup = reads.AbstractReadGroup(self.readGroupSet, 'notFound')
self.request.readGroupIds = [readGroup.getId()]
with self.assertRaises(exceptions.ReadGroupNotFoundException):
self.backend.readsGenerator(self.request)
def getReadGroup(self):
return reads.AbstractReadGroup(self.getReadGroupSet(), "readGroup")