def build_ctab_files(self,
ref_genome_path,
alignment_path,
output_dir,
num_threads=2):
"""
Runs tablemaker and generates the ctab files in the specified output_dir
:param ref_genome_path: File path to a reference genome or transcriptome assembly in GTF
format
:param alignment_path: File path to a reads alignment in BAM format
:param output_dir: directory to which the ctab files need to be written
:param num_threads: Number of prcessing threads (default=2)
:returns 0 if successful, else 1
"""
print('Running tablemaker...')
print "Args passed : ref_genome_path: {0} , alignment_path: {1} , output_dir: {2} , " \
"num_threads: {3} ".format(ref_genome_path, alignment_path, output_dir, num_threads)
tm_args = " -p {0} -o {1} -q -W -G {2} {3}".format(
str(num_threads), output_dir, ref_genome_path, alignment_path)
try:
print "Executing: tablemaker {0}".format(tm_args)
runProgram(logger=self.logger,
progName="tablemaker",
argStr=tm_args)
except Exception as ex:
log(
"Error executing tablemaker {0}. {1}".format(
tm_args, ex.message), logging.ERROR)
return 1
return 0
def get_stats(self, ifile, ipath):
"""
Generate simple statistics from a BAM file. The statistics collected include
counts of aligned and unaligned reads as well as all records with no start
coordinate.
:param ifile: bam file name
:param ipath: absolute path to bam file.
"""
if not ipath.startswith('/'):
raise Exception("Input path must be an absolute path. Provided: " +
str(ipath))
ifile = os.path.join(ipath, ifile)
# check if input file exists
if not os.path.exists(ifile):
raise RuntimeError(None,
'Input bam file does not exist: ' + str(ifile))
# get stats
self._check_prog()
try:
# samtools flagstat ifile
# samtools appears to operates on garbage-in-garbage out policy. i.e.
# it does not validate input and always returns True. Hence output
# value is not being checked.
stats = Popen('samtools flagstat {0}'.format(ifile),
shell=True,
stdin=PIPE,
stdout=PIPE)
stats, stderr = stats.communicate()
result = self._extractAlignmentStatsInfo(stats)
except Exception as ex:
log(
'failed to get stats from {0}'.format(ifile) + '. ' +
ex.message, logging.ERROR)
# return {} #TODO send back a dictionary with -1 values
return result
def validate(self,
ifile,
ipath,
ignore=[
'MATE_NOT_FOUND', 'MISSING_READ_GROUP',
'INVALID_MAPPING_QUALITY'
]):
"""
Validates the input bam file. Logs the errors if errors are found
:param ifile: bam file name
:param ipath: absolute path to bam file.
:param ignore: list of errors to ignore (see
http://broadinstitute.github.io/picard/command-line-overview.html#ValidateSamFile)
:returns 0 if successful, else 1
"""
if not ipath.startswith('/'):
raise Exception("Input path must be an absolute path. Provided: " +
str(ipath))
ifile = os.path.join(ipath, ifile)
# check if input file exists
if not os.path.exists(ifile):
raise RuntimeError(None,
'Input file does not exist: ' + str(ifile))
try:
# java -jar picard.jar ValidateSamFile I=ifile MODE=SUMMARY
validation = Popen(
'/usr/lib/jvm/java-8-oracle/bin/java -jar '
'/opt/picard/build/libs/picard.jar ValidateSamFile I={0} '
'MODE=SUMMARY'.format(ifile),
shell=True,
stdin=PIPE,
stdout=PIPE)
result, stderr = validation.communicate()
if self._is_valid(result, ignore):
log('{0} passed validation'.format(ifile), logging.INFO,
self.logger)
return 0
else:
log(
'{0} failed validation with errors: {1}'.format(
ifile, result), logging.ERROR, self.logger)
return 1
except Exception as ex:
log('{0} failed validation'.format(ifile) + '. ' + ex.message,
logging.ERROR, self.logger)
return 1
def create_bai_from_bam(self,
ifile,
ipath,
ofile=None,
opath=None,
validate=False,
ignore=[
'MATE_NOT_FOUND', 'MISSING_READ_GROUP',
'INVALID_MAPPING_QUALITY'
]):
"""
creates a bai file from a bam file
throws exceptions if input file is missing, invalid or if the output file could
not be written to disk
:param ifile: bam file name
:param ipath: absolute path to bam file
:param ofile: bai file name. If None, ifile name is used with the
extension '.bam' (if any) replaced with '.bai'
:param opath: path to bai file. If None, ipath will be used
:param validate: set to true if sam file needs to be validated. Default=False
:param ignore: see validate() method param
:returns 0 if successful, else 1
"""
# prepare input and output file paths
ifile, ofile, opath = self._prepare_paths(ifile, ipath, ofile, opath,
'.bam', '.bai')
# check if input file exists
if not os.path.exists(ifile):
raise RuntimeError(None,
'Input bam file does not exist: ' + str(ifile))
# validate input sam file
if validate and self.validate(ifile, ipath, ignore=ignore) == 1:
return 1
# convert
self._check_prog()
# samtools index ifile ofile
# samtools appears to operates on garbage-in-garbage out policy. i.e.
# it does not validate input and always returns True. Hence output
# value is not being checked.
try:
log('Creating bai from bam for file: ' + str(ifile) +
' with output file: ' + str(ofile) + ' and cwd: ' + str(opath))
create = Popen('samtools index {0} {1}'.format(ifile, ofile),
shell=True,
stdin=PIPE,
stdout=PIPE,
cwd=opath)
create.communicate()
except Exception as ex:
log(
'failed to convert {0} to {1}'.format(ifile, ofile) + '. ' +
ex.message, logging.ERROR)
return 1
return 0
def convert_sam_to_sorted_bam(self,
ifile,
ipath,
ofile=None,
opath=None,
validate=False,
ignore=[
'MATE_NOT_FOUND', 'MISSING_READ_GROUP',
'INVALID_MAPPING_QUALITY'
]):
"""
Converts the specified sam file to a sorted bam file.
throws exceptions if input file is missing, invalid or if the output file could
not be written to disk
:param ifile: sam file name
:param ipath: absolute path to sam file.
:param ofile: sorted bam file name. If None, ifile name is used with the
extension '.sam' (if any) replaced with '.bam'
:param opath: absolute path to sorted bam file. If None, ipath will be used
:param validate: set to true if sam file needs to be validated. Default=False
:param ignore: see validate() method param
:returns 0 if successful, else 1
"""
# prepare input and output file paths
ifile, ofile, opath = self._prepare_paths(ifile, ipath, ofile, opath,
'.sam', '.bam')
# check if input file exists
if not os.path.exists(ifile):
raise RuntimeError(None,
'Input sam file does not exist: ' + str(ifile))
# validate input sam file
if validate and self.validate(ifile, ipath, ignore=ignore) == 1:
return 1
# convert
self._check_prog()
# samtools view -bS ifile | samtools sort -l 9 -O BAM > ofile
# samtools appears to operates on garbage-in-garbage out policy. i.e.
# it does not validate input and always returns True. Hence output
# value is not being checked.
try:
log('Converting sam to sorted bam for file: ' + str(ifile) +
' with cwd: ' + str(opath))
sort = Popen('samtools sort -l 9 -O BAM > {0}'.format(ofile),
shell=True,
stdin=PIPE,
stdout=PIPE,
cwd=opath)
view = Popen('samtools view -bS {0}'.format(ifile),
shell=True,
stdout=sort.stdin,
cwd=opath)
result, stderr = sort.communicate(
) # samtools always returns success
view.wait()
except Exception as ex:
log(
'failed to convert {0} to {1}'.format(ifile, ofile) + '. ' +
ex.message, logging.ERROR)
return 0