def verifyFullConversion(self, readGroupSet, readGroup, reference):
"""
Verify that the conversion of the specified readGroup in the
specified readGroupSet for the specified reference is correct.
This involves pulling out the reads from the original BAM file
and comparing these with the converted SAM records.
"""
with tempfile.NamedTemporaryFile() as fileHandle:
converter = converters.SamConverter(self._client,
readGroup.getId(),
reference.getId(),
outputFileName=fileHandle.name)
converter.convert()
samFile = pysam.AlignmentFile(fileHandle.name, "r")
try:
convertedReads = list(samFile.fetch())
finally:
samFile.close()
samFile = pysam.AlignmentFile(readGroupSet.getSamFilePath(), "rb")
try:
sourceReads = []
referenceName = reference.getName().encode()
readGroupName = readGroup.getLocalId().encode()
for readAlignment in samFile.fetch(referenceName):
tags = dict(readAlignment.tags)
if 'RG' in tags and tags['RG'] == readGroupName:
sourceReads.append(readAlignment)
finally:
samFile.close()
self.verifySamRecordsEqual(sourceReads, convertedReads)
def run(self):
readGroup = self._httpClient.getReadGroup(self._readGroupIds[0])
iterator = self._httpClient.searchReads(
readGroupIds=self._readGroupIds, referenceId=self._referenceId,
start=self._start, end=self._end)
# do conversion
samConverter = converters.SamConverter(
readGroup, iterator, self._outputFile, self._binaryOutput)
samConverter.convert()
def run(self):
samConverter = converters.SamConverter(
self._client,
readGroupId=self._readGroupIds[0],
referenceId=self._referenceId,
start=self._start,
end=self._end,
outputFileName=self._outputFile,
binaryOutput=self._binaryOutput)
samConverter.convert()
def ga2sam_run(args):
# instantiate params
searchReadsRequest = RequestFactory(args).createSearchReadsRequest()
workarounds = getWorkarounds(args)
httpClient = client.HttpClient(args.baseUrl, args.verbose, workarounds,
args.key)
# do conversion
samConverter = converters.SamConverter(httpClient, searchReadsRequest,
args.outputFile, args.binaryOutput)
samConverter.convert()
def _testRoundTrip(self, binaryOutput):
with tempfile.NamedTemporaryFile() as fileHandle:
# write SAM file
filePath = fileHandle.name
samConverter = converters.SamConverter(None, self.getReads(),
filePath, binaryOutput)
samConverter.convert()
# read SAM file
samfile = pysam.AlignmentFile(filePath, "r")
reads = list(samfile.fetch())
self.assertEqual(reads[0].query_name, "SRR622461.77861202")
# TODO more in-depth testing
samfile.close()
def verifyFullConversion(self, readGroupSet, readGroup, reference):
"""
Verify that the conversion of the specified readGroup in the
specified readGroupSet for the specified reference is correct.
This involves pulling out the reads from the original BAM file
and comparing these with the converted SAM records.
"""
with tempfile.NamedTemporaryFile() as fileHandle:
converter = converters.SamConverter(self._client,
readGroup.getId(),
reference.getId(),
outputFileName=fileHandle.name)
converter.convert()
samFile = pysam.AlignmentFile(fileHandle.name, "r")
try:
# TODO suppressed because of pysam output:
# [W::sam_parse1] mapped mate cannot have zero coordinate;
# treated as unmapped
# and
# [W::sam_parse1] mapped mate cannot have zero coordinate;
# treated as unmapped
# see discussion in https://github.com/ga4gh/server/pull/789
with utils.suppressOutput():
convertedReads = list(samFile.fetch())
finally:
samFile.close()
samFile = pysam.AlignmentFile(readGroupSet.getDataUrl(), "rb")
try:
sourceReads = []
referenceName = reference.getName().encode()
readGroupName = readGroup.getLocalId().encode()
for readAlignment in samFile.fetch(referenceName):
tags = dict(readAlignment.tags)
if 'RG' in tags and tags['RG'] == readGroupName:
sourceReads.append(readAlignment)
finally:
samFile.close()
self.verifySamRecordsEqual(sourceReads, convertedReads)