def runEach(self,task_list):
### Call Cuffmerge function
used_tool = self.details['used_tool']
if used_tool == 'StringTie':
merged_gtf = rnaseq_util.call_stringtiemerge(diffexp_dir,merge_dir,num_threads,self.details['gtf_file'],assembly_file)
run_tool = "StringTie"
tool_version = "1.2.3"
elif used_tool == 'Cufflinks':
merged_gtf = rnaseq_util.call_cuffmerge(diffexp_dir,merge_dir,num_threads,gtf_file,assembly_file)
run_tool = "Tablemaker"
tool_version = '2.0.9'
cuffmerge_dir = os.path.join(self.directory,"cuffmerge")
merged_gtf = rnaseq_util.call_cuffmerge(self.directory,cuffmerge_dir,self.num_threads,self.details['gtf_file'],self.details['gtf_list_file'])
### Run DiffExpforBallgown
output_dir = os.path.join(self.directory,self.method_params['output_obj_name'])
diffexp_command = (' -p '+str(self.num_threads))
### Setting Advanced parameters for DiffExpforBallgown
if('time_series' in self.method_params and self.method_params['time_series'] != 0) : diffexp_command += (' -T ')
if('min_alignment_count' in self.method_params and self.method_params['min_alignment_count'] is not None ) : diffexp_command += (' -c '+str(self.method_params['min_alignment_count']))
if('multi_read_correct' in self.method_params and self.method_params['multi_read_correct'] != 0 ): diffexp_command += (' --multi-read-correct ')
if('library_type' in self.method_params and self.method_params['library_type'] is not None ) : diffexp_command += ( ' --library-type '+self.method_params['library_type'])
if('library_norm_method' in self.method_params and self.method_params['library_norm_method'] is not None ) : diffexp_command += ( ' --library-norm-method '+self.method_params['library_norm_method'])
try:
diffexp_command += " -o {0} -L {1} -u {2} {3}".format(output_dir,self.t_labels,merged_gtf,self.bam_files)
self.logger.info("Executing: diffexp {0}".format(diffexp_command))
ret = script_util.runProgram(None,"diffexp",diffexp_command,None,self.directory)
result = ret["result"]
#error = ret['stderr']
#print result
for line in result.splitlines(False):
self.logger.info(line)
stderr = ret["stderr"]
prev_value = ''
for line in stderr.splitlines(False):
if line.startswith('> Processing Locus'):
words = line.split()
cur_value = words[len(words) - 1]
if prev_value != cur_value:
prev_value = cur_value
self.logger.info(line)
else:
prev_value = ''
self.logger.info(line)
except Exception,e:
raise Exception(e)
raise Exception("Error executing diffexp {0},{1}".format(diffexp_command,e))
def runEach(self, task_list):
### Call Cuffmerge function
cuffmerge_dir = os.path.join(self.directory, "cuffmerge")
merged_gtf = rnaseq_util.call_cuffmerge(self.directory, cuffmerge_dir,
self.num_threads,
self.details['gtf_file'],
self.details['gtf_list_file'])
### Run Cuffdiff
output_dir = os.path.join(self.directory,
self.method_params['output_obj_name'])
cuffdiff_command = (' -p ' + str(self.num_threads))
### Setting Advanced parameters for Cuffdiff
if ('time_series' in self.method_params
and self.method_params['time_series'] != 0):
cuffdiff_command += (' -T ')
if ('min_alignment_count' in self.method_params
and self.method_params['min_alignment_count'] is not None):
cuffdiff_command += (
' -c ' + str(self.method_params['min_alignment_count']))
if ('multi_read_correct' in self.method_params
and self.method_params['multi_read_correct'] != 0):
cuffdiff_command += (' --multi-read-correct ')
if ('library_type' in self.method_params
and self.method_params['library_type'] is not None):
cuffdiff_command += (' --library-type ' +
self.method_params['library_type'])
if ('library_norm_method' in self.method_params
and self.method_params['library_norm_method'] is not None):
cuffdiff_command += (' --library-norm-method ' +
self.method_params['library_norm_method'])
try:
cuffdiff_command += " -o {0} -L {1} -u {2} {3}".format(
output_dir, self.t_labels, merged_gtf, self.bam_files)
self.logger.info(
"Executing: cuffdiff {0}".format(cuffdiff_command))
ret = script_util.runProgram(None, "cuffdiff", cuffdiff_command,
None, self.directory)
result = ret["result"]
for line in result.splitlines(False):
self.logger.info(line)
stderr = ret["stderr"]
prev_value = ''
for line in stderr.splitlines(False):
if line.startswith('> Processing Locus'):
words = line.split()
cur_value = words[len(words) - 1]
if prev_value != cur_value:
prev_value = cur_value
self.logger.info(line)
else:
prev_value = ''
self.logger.info(line)
except Exception, e:
raise Exception(e)
raise Exception("Error executing cuffdiff {0},{1}".format(
cuffdiff_command, e))
def prepare(self):
# for quick testing, we recover parameters here
ws_client = self.common_params['ws_client']
hs = self.common_params['hs_client']
params = self.method_params
logger = self.logger
token = self.common_params['user_token']
diffexp_dir = self.directory
self.details = rnaseq_util.get_details_for_diff_exp(logger,ws_client,hs,params['ws_id'],self.urls,diffexp_dir,params['expressionset_id'],token)
self.num_threads = mp.cpu_count()
self.num_jobs = 1
als = []
for l in self.details['labels']:
rep_files=[ (os.path.join(diffexp_dir+'/'+l,sub+'/accepted_hits.bam'), os.path.join(diffexp_dir+'/'+l,sub+'/transcripts.gtf')) for sub in os.listdir(os.path.join(diffexp_dir,l)) if os.path.isdir(os.path.join(diffexp_dir,l+'/'+sub))]
#rep_files=",".join([ os.path.join(diffexp_dir+'/'+l,sub+'/accepted_hits.bam') for sub in os.listdir(os.path.join(diffexp_dir,l)) if os.path.isdir(os.path.join(diffexp_dir,l+'/'+sub))])
als += rep_files
### Call Cuffmerge function
used_tool = self.details['used_tool']
merge_dir = os.path.join(diffexp_dir,"merge")
if used_tool == 'StringTie':
run_tool = "StringTie"
tool_version = "1.2.3"
merged_gtf = rnaseq_util.call_stringtiemerge(diffexp_dir,merge_dir,self.num_threads,self.details['gtf_file'],self.details['gtf_list_file'])
elif used_tool == 'Cufflinks':
merged_gtf = rnaseq_util.call_cuffmerge(diffexp_dir,merge_dir,num_threads,gtf_file,self.details['gtf_list_file'])
run_tool = "Tablemaker"
tool_version = '2.0.9'
merged_gtf = rnaseq_util.call_cuffmerge(diffexp_dir,merge_dir,self.num_threads,self.details['gtf_file'],self.details['gtf_list_file'])
self.bam_files = " ".join([i for i in als])
self.t_labels = ",".join(self.details['labels'])
ballgown_dir = os.path.join(diffexp_dir,"ballgown")
if not os.path.exists(ballgown_dir): os.mkdir(ballgown_dir)
### Make Input_dir from expression_file_name
self.task_list = [self.__class__]
def runEach(self,task_list):
### Call Cuffmerge function
cuffmerge_dir = os.path.join(self.directory,"cuffmerge")
merged_gtf = rnaseq_util.call_cuffmerge(self.directory,cuffmerge_dir,self.num_threads,self.details['gtf_file'],self.details['gtf_list_file'])
### Run Cuffdiff
output_dir = os.path.join(self.directory,self.method_params['output_obj_name'])
cuffdiff_command = (' -p '+str(self.num_threads))
### Setting Advanced parameters for Cuffdiff
if('time_series' in self.method_params and self.method_params['time_series'] != 0) : cuffdiff_command += (' -T ')
if('min_alignment_count' in self.method_params and self.method_params['min_alignment_count'] is not None ) : cuffdiff_command += (' -c '+str(self.method_params['min_alignment_count']))
if('multi_read_correct' in self.method_params and self.method_params['multi_read_correct'] != 0 ): cuffdiff_command += (' --multi-read-correct ')
if('library_type' in self.method_params and self.method_params['library_type'] is not None ) : cuffdiff_command += ( ' --library-type '+self.method_params['library_type'])
if('library_norm_method' in self.method_params and self.method_params['library_norm_method'] is not None ) : cuffdiff_command += ( ' --library-norm-method '+self.method_params['library_norm_method'])
try:
cuffdiff_command += " -o {0} -L {1} -u {2} {3}".format(output_dir,self.t_labels,merged_gtf,self.bam_files)
self.logger.info("Executing: cuffdiff {0}".format(cuffdiff_command))
ret = script_util.runProgram(None,"cuffdiff",cuffdiff_command,None,self.directory)
result = ret["result"]
for line in result.splitlines(False):
self.logger.info(line)
stderr = ret["stderr"]
prev_value = ''
for line in stderr.splitlines(False):
if line.startswith('> Processing Locus'):
words = line.split()
cur_value = words[len(words) - 1]
if prev_value != cur_value:
prev_value = cur_value
self.logger.info(line)
else:
prev_value = ''
self.logger.info(line)
except Exception,e:
raise Exception(e)
raise Exception("Error executing cuffdiff {0},{1}".format(cuffdiff_command,e))
def runEach(self, task_list):
logger = self.logger
### Call Cuffmerge function
used_tool = self.details['used_tool']
logger.info('in DiffExpfoBallgown.runEach()')
if used_tool == "Ballgown (Bioconductor)":
#merged_gtf = rnaseq_util.call_stringtiemerge(diffexp_dir,merge_dir,num_threads,self.details['gtf_file'],assembly_file)
#run_tool = "StringTie"
#tool_version = "1.2.3"
# For now, take no action for StringTie processing
logger.info('Exiting immediately - StringTie case')
return
elif used_tool == 'Cufflinks':
merged_gtf = rnaseq_util.call_cuffmerge(diffexp_dir, merge_dir,
num_threads, gtf_file,
assembly_file)
run_tool = "Tablemaker"
tool_version = '2.0.9'
cuffmerge_dir = os.path.join(self.directory, "cuffmerge")
merged_gtf = rnaseq_util.call_cuffmerge(self.directory, cuffmerge_dir,
self.num_threads,
self.details['gtf_file'],
self.details['gtf_list_file'])
### Run DiffExpforBallgown
output_dir = os.path.join(self.directory,
self.method_params['output_obj_name'])
diffexp_command = (' -p ' + str(self.num_threads))
### Setting Advanced parameters for DiffExpforBallgown
if ('time_series' in self.method_params
and self.method_params['time_series'] != 0):
diffexp_command += (' -T ')
if ('min_alignment_count' in self.method_params
and self.method_params['min_alignment_count'] is not None):
diffexp_command += (' -c ' +
str(self.method_params['min_alignment_count']))
if ('multi_read_correct' in self.method_params
and self.method_params['multi_read_correct'] != 0):
diffexp_command += (' --multi-read-correct ')
if ('library_type' in self.method_params
and self.method_params['library_type'] is not None):
diffexp_command += (' --library-type ' +
self.method_params['library_type'])
if ('library_norm_method' in self.method_params
and self.method_params['library_norm_method'] is not None):
diffexp_command += (' --library-norm-method ' +
self.method_params['library_norm_method'])
try:
diffexp_command += " -o {0} -L {1} -u {2} {3}".format(
output_dir, self.t_labels, merged_gtf, self.bam_files)
logger.info("Executing: diffexp {0}".format(diffexp_command))
ret = script_util.runProgram(None, "diffexp", diffexp_command,
None, self.directory)
result = ret["result"]
#error = ret['stderr']
#print result
for line in result.splitlines(False):
logger.info(line)
stderr = ret["stderr"]
prev_value = ''
for line in stderr.splitlines(False):
if line.startswith('> Processing Locus'):
words = line.split()
cur_value = words[len(words) - 1]
if prev_value != cur_value:
prev_value = cur_value
logger.info(line)
else:
prev_value = ''
logger.info(line)
except Exception, e:
raise Exception(e)
raise Exception("Error executing diffexp {0},{1}".format(
diffexp_command, e))
