def convert(self):
# set flags
if self.inputFileFormat == AlignmentFileConstants.SAM:
inputFlags = "r"
elif self.inputFileFormat == AlignmentFileConstants.BAM:
inputFlags = "rb"
if self.outputFileFormat == AlignmentFileConstants.SAM:
outputFlags = "wh"
elif self.outputFileFormat == AlignmentFileConstants.BAM:
outputFlags = "wb"
# open files
inputFile = pysam.AlignmentFile(self.args.inputFile, inputFlags)
outputFile = pysam.AlignmentFile(self.args.outputFile,
outputFlags,
header=inputFile.header)
outputFilePath = outputFile.filename
utils.log("Creating alignment file '{}'".format(outputFilePath))
# write new file
for _ in xrange(self.args.numLines):
alignedSegment = inputFile.next()
outputFile.write(alignedSegment)
# clean up
inputFile.close()
outputFile.close()
# create index file
if (not self.args.skipIndexing
and self.outputFileFormat == AlignmentFileConstants.BAM):
indexFilePath = "{}.{}".format(outputFilePath,
AlignmentFileConstants.BAI.lower())
utils.log("Creating index file '{}'".format(indexFilePath))
pysam.index(outputFilePath)
def convert(self):
# set flags
if self.inputFileFormat == AlignmentFileConstants.SAM:
inputFlags = "r"
elif self.inputFileFormat == AlignmentFileConstants.BAM:
inputFlags = "rb"
if self.outputFileFormat == AlignmentFileConstants.SAM:
outputFlags = "wh"
elif self.outputFileFormat == AlignmentFileConstants.BAM:
outputFlags = "wb"
# open files
inputFile = pysam.AlignmentFile(
self.args.inputFile, inputFlags)
outputFile = pysam.AlignmentFile(
self.args.outputFile, outputFlags, header=inputFile.header)
outputFilePath = outputFile.filename
utils.log("Creating alignment file '{}'".format(outputFilePath))
# write new file
for _ in xrange(self.args.numLines):
alignedSegment = inputFile.next()
outputFile.write(alignedSegment)
# clean up
inputFile.close()
outputFile.close()
# create index file
if (not self.args.skipIndexing and
self.outputFileFormat == AlignmentFileConstants.BAM):
indexFilePath = "{}.{}".format(
outputFilePath, AlignmentFileConstants.BAI.lower())
utils.log("Creating index file '{}'".format(indexFilePath))
pysam.index(outputFilePath)
def testLog(self):
utils.log("message")
self.assertEquals(self.printMock.call_count, 1)
def __init__(self, inputDirectory, outputDirectory, force):
"""
Converts human readable dataset from compliance repository,
and translates it into a reference-server readable filesystem
with binary files.
:param inputDirectory: location of
the human readable compliance dataset
:param outputDirectory: location of
the file hierarchy suitable for deploying on the reference server
"""
self.inputDirectory = inputDirectory
self.outputDirectory = outputDirectory
self.repoPath = os.path.abspath(
os.path.join(outputDirectory, "registry.db"))
self.tempdir = None
if os.path.exists(self.outputDirectory):
if force:
utils.log("Removing existing output directory at '{}'".format(
self.outputDirectory))
shutil.rmtree(self.outputDirectory)
else:
utils.log("Output directory '{}' already exists".format(
self.outputDirectory))
utils.log("Please specify an output path that does not exist")
utils.log("Exiting...")
exit(1)
# If no input directory is specified download from GitHub
if inputDirectory is None:
utils.log("Downloading test data...")
self.tempdir = tempfile.mkdtemp()
assert (os.path.exists(self.tempdir))
url = "https://github.com/ga4gh/compliance/archive/master.zip"
filePath = os.path.join(self.tempdir, 'compliance-master.zip')
downloader = file_downloader.HttpFileDownloader(url, filePath)
downloader.download()
utils.log("Extracting test data...")
with zipfile.ZipFile(filePath, "r") as z:
z.extractall(self.tempdir)
self.inputDirectory = os.path.join(self.tempdir,
'compliance-master',
'test-data')
repo = datarepo.SqlDataRepository(self.repoPath)
self.repo = repo
def cleanup(self):
if self.tempdir is not None:
shutil.rmtree(self.tempdir)
utils.log("Done converting compliance data.")
utils.log("Result in '{}'".format(self.outputDirectory))
def runCommandCheckWarnings(self, cmd):
utils.log("Running '{}'".format(cmd))
splits = shlex.split(cmd)
output = subprocess.check_output(splits).split('\n')
self.ensureNoWarnings(output, cmd)
def __init__(self, inputDirectory, outputDirectory, force):
"""
Converts human readable dataset from compliance repository,
and translates it into a reference-server readable filesystem
with binary files.
:param inputDirectory: location of
the human readable compliance dataset
:param outputDirectory: location of
the file hierarchy suitable for deploying on the reference server
"""
self.inputDirectory = inputDirectory
self.outputDirectory = outputDirectory
self.repoPath = os.path.abspath(
os.path.join(outputDirectory, "registry.db"))
self.tempdir = None
if os.path.exists(self.outputDirectory):
if force:
utils.log(
"Removing existing output directory at '{}'".format(
self.outputDirectory))
shutil.rmtree(self.outputDirectory)
else:
utils.log(
"Output directory '{}' already exists".format(
self.outputDirectory))
utils.log(
"Please specify an output path that does not exist")
utils.log("Exiting...")
exit(1)
# If no input directory is specified download from GitHub
if inputDirectory is None:
utils.log("Downloading test data...")
self.tempdir = tempfile.mkdtemp()
assert(os.path.exists(self.tempdir))
url = "https://github.com/ga4gh/compliance/archive/master.zip"
filePath = os.path.join(self.tempdir, 'compliance-master.zip')
downloader = file_downloader.HttpFileDownloader(url, filePath)
downloader.download()
utils.log("Extracting test data...")
with zipfile.ZipFile(filePath, "r") as z:
z.extractall(self.tempdir)
self.inputDirectory = os.path.join(
self.tempdir, 'compliance-master', 'test-data')
repo = datarepo.SqlDataRepository(self.repoPath)
self.repo = repo
def cleanup(self):
if self.tempdir is not None:
shutil.rmtree(self.tempdir)
utils.log("Done converting compliance data.")
utils.log("Result in '{}'".format(self.outputDirectory))
def log(self, logStr):
utils.log("{0} {1}".format(self.logStrPrefix, logStr))
