def _CallHisat2(logger,services,ws_client,hs,ws_id,sample_type,num_threads,read_sample,condition,directory,genome_id,sampleset_id,params,token):
#logger.info("Downloading Read Sample{0}".format(read_sample))
print "Downloading Read Sample{0}".format(read_sample)
if not logger:
logger = handler_util.create_logger(directory,"run_Hisat2_"+read_sample)
logger.info("Downloading Read Sample{0}".format(read_sample))
try:
r_sample = ws_client.get_objects(
[{ 'name' : read_sample, 'workspace' : ws_id}])[0]
r_sample_info = ws_client.get_object_info_new({"objects": [{'name': read_sample, 'workspace': ws_id}]})[0]
sample_type = r_sample_info[2].split('-')[0]
input_direc = os.path.join(directory,read_sample.split('.')[0]+"_hisat2_input")
if not os.path.exists(input_direc): os.mkdir(input_direc)
output_name = read_sample.split('.')[0]+"_hisat2_alignment"
output_dir = os.path.join(directory,output_name)
if not os.path.exists(output_dir): os.mkdir(output_dir)
hisat2_base =os.path.join(directory,handler_util.get_file_with_suffix(directory,".1.ht2"))
### Adding advanced options to Bowtie2Call
hisat2_cmd = ''
hisat2_cmd += ( ' -p {0}'.format(num_threads))
if('quality_score' in params and params['quality_score'] is not None): hisat2_cmd += ( ' --'+params['quality_score'])
if('alignment_type' in params and params['alignment_type'] is not None): hisat2_cmd += ( ' --'+params['alignment_type'] )
if('trim5' in params and params['trim5'] is not None): hisat2_cmd += ( ' --trim5 '+str(params['trim5']))
if('trim3' in params and params['trim3'] is not None): hisat2_cmd += ( ' --trim3 '+str(params['trim3']))
if('np' in params and params['np'] is not None): hisat2_cmd += ( ' --np '+str(params['np']))
if('minins' in params and params['minins'] is not None): hisat2_cmd += ( ' --minins '+str(params['minins']))
if('maxins' in params and params['maxins'] is not None): hisat2_cmd += ( ' --maxins '+str(params['maxins']))
#if('orientation' in params and params['orientation'] is not None): hisat2_cmd += ( ' --'+params['orientation'])
if('min_intron_length' in params and params['min_intron_length'] is not None): hisat2_cmd += ( ' --min-intronlen '+str(params['min_intron_length']))
if('max_intron_length' in params and params['max_intron_length'] is not None): hisat2_cmd += ( ' --max-intronlen '+str(params['max_intron_length']))
if('no_spliced_alignment' in params and params['no_spliced_alignment'] != 0): hisat2_cmd += ( ' --no-spliced-alignment')
if('transcriptome_mapping_only' in params and params['transcriptome_mapping_only'] != 0): hisat2_cmd += ( ' --transcriptome-mapping-only')
if('tailor_alignments' in params and params['tailor_alignments'] is not None):
hisat2_cmd += ( ' --'+params['tailor_alignments'])
out_file = output_dir +"/accepted_hits.sam"
if sample_type == 'KBaseAssembly.SingleEndLibrary':
lib_type = 'SingleEnd'
read_id = r_sample['data']['handle']['id']
read_name = r_sample['data']['handle']['file_name']
try:
script_util.download_file_from_shock(logger, shock_service_url=services['shock_service_url'], shock_id=read_id,filename=read_name, directory=input_direc,token=token)
hisat2_cmd += " -U {0} -x {1} -S {2}".format(os.path.join(input_direc,read_name),hisat2_base,out_file)
except Exception,e:
#logger.exception( "Unable to download shock file , {0}".format(read_name))
raise Exception( "Unable to download shock file , {0}".format(read_name))
if sample_type == 'KBaseAssembly.PairedEndLibrary':
lib_type = 'PairedEnd'
if('orientation' in params and params['orientation'] is not None): hisat2_cmd += ( ' --'+params['orientation'])
read1_id = r_sample['data']['handle_1']['id']
read1_name = r_sample['data']['handle_1']['file_name']
read2_id = r_sample['data']['handle_2']['id']
read2_name = r_sample['data']['handle_2']['file_name']
try:
script_util.download_file_from_shock(logger, shock_service_url=services['shock_service_url'], shock_id=read1_id,filename=read1_name, directory=input_direc,token=token)
script_util.download_file_from_shock(logger, shock_service_url=services['shock_service_url'], shock_id=read2_id,filename=read2_name, directory=input_direc,token=token)
hisat2_cmd += " -1 {0} -2 {1} -x {2} -S {3}".format(os.path.join(input_direc,read1_name),os.path.join(output_dir,read2_name),hisat2_base,out_file)
except Exception,e:
#logger.Exception( "Unable to download shock file , {0} or {1}".format(read1_name,read2_name))
raise Exception( "Unable to download shock file , {0} or {1}".format(read1_name,read2_name))
def _CallBowtie2(logger,services,ws_client,hs,ws_id,sample_type,num_threads,read_sample,condition,directory,bowtie2index_id,genome_id,sampleset_id,params,token):
#logger.info("Downloading Read Sample{0}".format(read_sample))
print "Downloading Read Sample{0}".format(read_sample)
if not logger:
logger = create_logger(directory,"run_Bowtie2_"+read_sample)
logger.info("Downloading Read Sample{0}".format(read_sample))
try:
r_sample = ws_client.get_objects(
[{ 'name' : read_sample, 'workspace' : ws_id}])[0]
r_sample_info = ws_client.get_object_info_new({"objects": [{'name': read_sample, 'workspace': ws_id}]})[0]
sample_type = r_sample_info[2].split('-')[0]
output_name = read_sample.split('.')[0]+"_bowtie2_alignment"
output_dir = os.path.join(directory,output_name)
if not os.path.exists(output_dir): os.mkdir(output_dir)
out_file = output_dir +"/accepted_hits.sam"
bowtie2_base =os.path.join(directory,handler_util.get_file_with_suffix(directory,".rev.1.bt2"))
### Adding advanced options to Bowtie2Call
bowtie2_cmd = ''
bowtie2_cmd += ( ' -p {0}'.format(num_threads))
if('quality_score' in params and params['quality_score'] is not None): bowtie2_cmd += ( ' --'+params['quality_score'])
if('alignment_type' in params and params['alignment_type'] is not None): bowtie2_cmd += ( ' --'+params['alignment_type'] )
if('preset_options' in params and params['preset_options'] is not None ) and ('alignment_type' in params and params['alignment_type'] is not None):
if (params['alignment_type'] == 'local'):
bowtie2_cmd += (' --'+params['preset_options']+'-local')
else: bowtie2_cmd += (' --'+params['preset_options'] )
if('trim5' in params and params['trim5'] is not None): bowtie2_cmd += ( ' --trim5 '+str(params['trim5']))
if('trim3' in params and params['trim3'] is not None): bowtie2_cmd += ( ' --trim3 '+str(params['trim3']))
if('np' in params and params['np'] is not None): bowtie2_cmd += ( ' --np '+str(params['np']))
if('minins' in params and params['minins'] is not None): bowtie2_cmd += ( ' --minins '+str(params['minins']))
if('maxins' in params and params['maxins'] is not None): bowtie2_cmd += ( ' --maxins '+str(params['maxins']))
if('orientation' in params and params['orientation'] is not None): bowtie2_cmd += ( ' --'+params['orientation'])
if sample_type == 'KBaseAssembly.SingleEndLibrary':
lib_type = 'SingleEnd'
read_id = r_sample['data']['handle']['id']
read_name = r_sample['data']['handle']['file_name']
try:
script_util.download_file_from_shock(logger, shock_service_url=services['shock_service_url'], shock_id=read_id,filename=read_name, directory=directory,token=token)
bowtie2_cmd += " -U {0} -x {1} -S {2}".format(os.path.join(directory,read_name),bowtie2_base,out_file)
except Exception,e:
#logger.exception( "Unable to download shock file , {0}".format(read_name))
raise Exception( "Unable to download shock file , {0}".format(read_name))
if sample_type == 'KBaseAssembly.PairedEndLibrary':
lib_type = 'PairedEnd'
read1_id = r_sample['data']['handle_1']['id']
read1_name = r_sample['data']['handle_1']['file_name']
read2_id = r_sample['data']['handle_2']['id']
read2_name = r_sample['data']['handle_2']['file_name']
try:
script_util.download_file_from_shock(logger, shock_service_url=services['shock_service_url'], shock_id=read1_id,filename=read1_name, directory=directory,token=token)
script_util.download_file_from_shock(logger, shock_service_url=services['shock_service_url'], shock_id=read2_id,filename=read2_name, directory=directory,token=token)
bowtie2_cmd += " -1 {0} -2 {1} -x {2} -S {3}".format(os.path.join(directory,read1_name),os.path.join(directory,read2_name),bowtie2_base,out_file)
except Exception,e:
#logger.Exception( "Unable to download shock file , {0} or {1}".format(read1_name,read2_name))
raise Exception( "Unable to download shock file , {0} or {1}".format(read1_name,read2_name))
def _CallTophat(logger,services,ws_client,hs,ws_id,sample_type,num_threads,read_sample,gtf_file,condition,directory,bowtie2index_id,genome_id,sampleset_id,params,token):
print "Downloading Read Sample{0}".format(read_sample)
if not logger:
logger = create_logger(directory,"run_Tophat_"+read_sample)
try:
r_sample = ws_client.get_objects(
[{ 'name' : read_sample, 'workspace' : ws_id}])[0]
r_sample_info = ws_client.get_object_info_new({"objects": [{'name': read_sample, 'workspace': ws_id}]})[0]
sample_type = r_sample_info[2].split('-')[0]
output_name = read_sample.split('.')[0]+"_tophat_alignment"
output_dir = os.path.join(directory,output_name)
#if not os.path.exists(output_dir): os.makedirs(output_dir)
#out_file = output_dir +"/accepted_hits.sam"
bowtie2_base =os.path.join(directory,handler_util.get_file_with_suffix(directory,".rev.1.bt2"))
### Adding advanced options to tophat command
tophat_cmd = (' -p '+str(num_threads))
if('max_intron_length' in params and params['max_intron_length'] is not None ) : tophat_cmd += (' -I '+str(params['max_intron_length']))
if('min_intron_length' in params and params['min_intron_length'] is not None ): tophat_cmd += (' -i '+str(params['min_intron_length']))
if('min_anchor_length' in params and params['min_anchor_length'] is not None ): tophat_cmd += (' -a '+str(params['min_anchor_length']))
if('read_edit_dist' in params and params['read_edit_dist'] is not None ) : tophat_cmd += (' --read-edit-dist '+str(params['read_edit_dist']))
if('read_gap_length' in params and params['read_gap_length'] is not None) : tophat_cmd += (' --read-gap-length '+str(params['read_gap_length']))
if('read_mismatches' in params and params['read_mismatches'] is not None) : tophat_cmd += (' -N '+str(params['read_mismatches']))
if('library_type' in params and params['library_type'] is not None ) : tophat_cmd += (' --library-type ' + params['library_type'])
if('report_secondary_alignments' in params and int(params['report_secondary_alignments']) == 1) : tophat_cmd += ' --report-secondary-alignments'
if('no_coverage_search' in params and int(params['no_coverage_search']) == 1): tophat_cmd += ' --no-coverage-search'
if('preset_options' in params and params['preset_options'] is not None ): tophat_cmd += ' --'+params['preset_options']
if sample_type == 'KBaseAssembly.SingleEndLibrary':
lib_type = 'SingleEnd'
read_id = r_sample['data']['handle']['id']
read_name = r_sample['data']['handle']['file_name']
try:
script_util.download_file_from_shock(logger, shock_service_url=services['shock_service_url'], shock_id=read_id,filename=read_name, directory=directory,token=token)
tophat_cmd += ' -o {0} -G {1} {2} {3}'.format(output_dir,gtf_file,bowtie2_base,os.path.join(directory,read_name))
except Exception,e:
raise Exception( "Unable to download shock file , {0}".format(read_name))
if sample_type == 'KBaseAssembly.PairedEndLibrary':
lib_type = 'PairedEnd'
read1_id = r_sample['data']['handle_1']['id']
read1_name = r_sample['data']['handle_1']['file_name']
read2_id = r_sample['data']['handle_2']['id']
read2_name = r_sample['data']['handle_2']['file_name']
try:
script_util.download_file_from_shock(logger, shock_service_url=services['shock_service_url'], shock_id=read1_id,filename=read1_name, directory=directory,token=token)
script_util.download_file_from_shock(logger, shock_service_url=services['shock_service_url'], shock_id=read2_id,filename=read2_name, directory=directory,token=token)
tophat_cmd += ' -o {0} -G {1} {2} {3} {4}'.format(output_dir,gtf_file,bowtie2_base,os.path.join(directory,read1_name),os.path.join(directory,read2_name))
except Exception,e:
raise Exception( "Unable to download shock file , {0} or {1}".format(read1_name,read2_name))
def _CallHisat2(logger, services, ws_client, hs, ws_id, sample_type,
num_threads, read_sample, condition, directory, genome_id,
sampleset_id, params, token):
#logger.info("Downloading Read Sample{0}".format(read_sample))
print "Downloading Read Sample{0}".format(read_sample)
if not logger:
logger = handler_util.create_logger(directory,
"run_Hisat2_" + read_sample)
logger.info("Downloading Read Sample{0}".format(read_sample))
try:
r_sample = ws_client.get_objects([{
'name': read_sample,
'workspace': ws_id
}])[0]
r_sample_info = ws_client.get_object_info_new(
{"objects": [{
'name': read_sample,
'workspace': ws_id
}]})[0]
sample_type = r_sample_info[2].split('-')[0]
input_direc = os.path.join(directory,
read_sample.split('.')[0] + "_hisat2_input")
if not os.path.exists(input_direc): os.mkdir(input_direc)
output_name = read_sample.split('.')[0] + "_hisat2_alignment"
output_dir = os.path.join(directory, output_name)
if not os.path.exists(output_dir): os.mkdir(output_dir)
hisat2_base = os.path.join(
directory, handler_util.get_file_with_suffix(directory, ".1.ht2"))
### Adding advanced options to Bowtie2Call
hisat2_cmd = ''
hisat2_cmd += (' -p {0}'.format(num_threads))
if ('quality_score' in params and params['quality_score'] is not None):
hisat2_cmd += (' --' + params['quality_score'])
if ('alignment_type' in params
and params['alignment_type'] is not None):
hisat2_cmd += (' --' + params['alignment_type'])
if ('trim5' in params and params['trim5'] is not None):
hisat2_cmd += (' --trim5 ' + str(params['trim5']))
if ('trim3' in params and params['trim3'] is not None):
hisat2_cmd += (' --trim3 ' + str(params['trim3']))
if ('np' in params and params['np'] is not None):
hisat2_cmd += (' --np ' + str(params['np']))
if ('minins' in params and params['minins'] is not None):
hisat2_cmd += (' --minins ' + str(params['minins']))
if ('maxins' in params and params['maxins'] is not None):
hisat2_cmd += (' --maxins ' + str(params['maxins']))
#if('orientation' in params and params['orientation'] is not None): hisat2_cmd += ( ' --'+params['orientation'])
if ('min_intron_length' in params
and params['min_intron_length'] is not None):
hisat2_cmd += (' --min-intronlen ' +
str(params['min_intron_length']))
if ('max_intron_length' in params
and params['max_intron_length'] is not None):
hisat2_cmd += (' --max-intronlen ' +
str(params['max_intron_length']))
if ('no_spliced_alignment' in params
and params['no_spliced_alignment'] != 0):
hisat2_cmd += (' --no-spliced-alignment')
if ('transcriptome_mapping_only' in params
and params['transcriptome_mapping_only'] != 0):
hisat2_cmd += (' --transcriptome-mapping-only')
if ('tailor_alignments' in params
and params['tailor_alignments'] is not None):
hisat2_cmd += (' --' + params['tailor_alignments'])
out_file = output_dir + "/accepted_hits.sam"
if sample_type == 'KBaseAssembly.SingleEndLibrary':
lib_type = 'SingleEnd'
read_id = r_sample['data']['handle']['id']
read_name = r_sample['data']['handle']['file_name']
try:
script_util.download_file_from_shock(
logger,
shock_service_url=services['shock_service_url'],
shock_id=read_id,
filename=read_name,
directory=input_direc,
token=token)
hisat2_cmd += " -U {0} -x {1} -S {2}".format(
os.path.join(input_direc, read_name), hisat2_base,
out_file)
except Exception, e:
#logger.exception( "Unable to download shock file , {0}".format(read_name))
raise Exception(
"Unable to download shock file , {0}".format(read_name))
if sample_type == 'KBaseAssembly.PairedEndLibrary':
lib_type = 'PairedEnd'
if ('orientation' in params and params['orientation'] is not None):
hisat2_cmd += (' --' + params['orientation'])
read1_id = r_sample['data']['handle_1']['id']
read1_name = r_sample['data']['handle_1']['file_name']
read2_id = r_sample['data']['handle_2']['id']
read2_name = r_sample['data']['handle_2']['file_name']
try:
script_util.download_file_from_shock(
logger,
shock_service_url=services['shock_service_url'],
shock_id=read1_id,
filename=read1_name,
directory=input_direc,
token=token)
script_util.download_file_from_shock(
logger,
shock_service_url=services['shock_service_url'],
shock_id=read2_id,
filename=read2_name,
directory=input_direc,
token=token)
hisat2_cmd += " -1 {0} -2 {1} -x {2} -S {3}".format(
os.path.join(input_direc, read1_name),
os.path.join(output_dir, read2_name), hisat2_base,
out_file)
except Exception, e:
#logger.Exception( "Unable to download shock file , {0} or {1}".format(read1_name,read2_name))
raise Exception(
"Unable to download shock file , {0} or {1}".format(
read1_name, read2_name))
raise Bowtie2SampleException('Either of the Library typed objects SingleEndLibrary or PairedEndLibrary is required')
r_label = 'Single'
### Get the Bw index file
bw_index_files = script_util.check_and_download_existing_handle_obj(logger,ws_client,self.urls,params['ws_id'],params['bowtie_index'],"KBaseRNASeq.Bowtie2Indexes",bowtie2_dir,token)
try:
logger.info("Unzipping Bowtie2 Indices")
script_util.unzip_files(logger,os.path.join(bowtie2_dir,bw_index_files),bowtie2_dir)
mv_dir= handler_util.get_dir(bowtie2_dir)
if mv_dir is not None:
script_util.move_files(logger,mv_dir,bowtie2_dir)
except Exception, e:
logger.error("".join(traceback.format_exc()))
raise Exception("Unzip indexfile error: Please contact [email protected]")
### Build Index for the fasta file
fasta_file =os.path.join(bowtie2_dir,handler_util.get_file_with_suffix(bowtie2_dir,".fa")+".fa")
bowtie2base =os.path.join(bowtie2_dir,handler_util.get_file_with_suffix(bowtie2_dir,".fa"))
bowtie2base_cmd = '{0} {1}'.format(fasta_file,bowtie2base)
try:
logger.info("Building Index for Hisat2 {0}".format(bowtie2base_cmd))
cmdline_output = script_util.runProgram(logger,"bowtie2-build",bowtie2base_cmd,None,bowtie2_dir)
except Exception,e:
raise Exception("Failed to run command {0}".format(bowtie2base_cmd))
### Check if GTF object exists in the workspace pull the gtf
ref_id = bowtie_index['data']['genome_id']
genome_name = ws_client.get_object_info_new({"objects": [{'ref' : ref_id }] })[0][1]
ws_gtf = genome_name+"_GTF"
gtf_file = script_util.check_and_download_existing_handle_obj(logger,ws_client,self.urls,params['ws_id'],ws_gtf,"KBaseRNASeq.GFFAnnotation",bowtie2_dir,token)
if gtf_file is None:
rnaseq_util.create_gtf_annotation_from_genome(logger,ws_client,hs,self.urls,params['ws_id'],ref_id,genome_name,bowtie2_dir,token)
# Determine the num_threads provided by the user otherwise default the number of threads to 2
.format(",".join(missing_objs), params['ws_id']))
### Build Hisat2 index
fasta_file = script_util.generate_fasta(logger, services, token,
annotation_id, hisat2_dir,
params['genome_id'])
logger.info("Sanitizing the fasta file to correct id names {}".format(
datetime.datetime.utcnow()))
mapping_filename = c_mapping.create_sanitized_contig_ids(fasta_file)
c_mapping.replace_fasta_contig_ids(fasta_file,
mapping_filename,
to_modified=True)
logger.info("Generating FASTA file completed successfully : {}".format(
datetime.datetime.utcnow()))
hisat2base = os.path.join(
hisat2_dir, handler_util.get_file_with_suffix(hisat2_dir, ".fasta"))
hisat2base_cmd = '{0} {1}'.format(fasta_file, hisat2base)
try:
logger.info("Building Index for Hisat2 {0}".format(hisat2base_cmd))
cmdline_output = script_util.runProgram(logger, "hisat2-build",
hisat2base_cmd, None,
hisat2_dir)
except Exception, e:
raise Exception("Failed to run command {0}".format(hisat2base_cmd))
ws_gtf = params['genome_id'] + "_GTF"
ret = script_util.if_obj_exists(None, ws_client, params['ws_id'],
"KBaseRNASeq.GFFAnnotation", [ws_gtf])
print ret
if not ret is None:
logger.info(
"GFF Annotation Exist for Genome Annotation {0}.... Skipping step "
def _CallBowtie2(
logger,
services,
ws_client,
hs,
ws_id,
sample_type,
num_threads,
read_sample,
condition,
directory,
bowtie2index_id,
genome_id,
sampleset_id,
params,
token,
):
# logger.info("Downloading Read Sample{0}".format(read_sample))
print "Downloading Read Sample{0}".format(read_sample)
if not logger:
logger = create_logger(directory, "run_Bowtie2_" + read_sample)
logger.info("Downloading Read Sample{0}".format(read_sample))
try:
r_sample = ws_client.get_objects([{"name": read_sample, "workspace": ws_id}])[0]
r_sample_info = ws_client.get_object_info_new({"objects": [{"name": read_sample, "workspace": ws_id}]})[0]
sample_type = r_sample_info[2].split("-")[0]
output_name = read_sample.split(".")[0] + "_bowtie2_alignment"
output_dir = os.path.join(directory, output_name)
if not os.path.exists(output_dir):
os.mkdir(output_dir)
out_file = output_dir + "/accepted_hits.sam"
bowtie2_base = os.path.join(directory, handler_util.get_file_with_suffix(directory, ".rev.1.bt2"))
### Adding advanced options to Bowtie2Call
bowtie2_cmd = ""
bowtie2_cmd += " -p {0}".format(num_threads)
if "quality_score" in params and params["quality_score"] is not None:
bowtie2_cmd += " --" + params["quality_score"]
if "alignment_type" in params and params["alignment_type"] is not None:
bowtie2_cmd += " --" + params["alignment_type"]
if ("preset_options" in params and params["preset_options"] is not None) and (
"alignment_type" in params and params["alignment_type"] is not None
):
if params["alignment_type"] == "local":
bowtie2_cmd += " --" + params["preset_options"] + "-local"
else:
bowtie2_cmd += " --" + params["preset_options"]
if "trim5" in params and params["trim5"] is not None:
bowtie2_cmd += " --trim5 " + str(params["trim5"])
if "trim3" in params and params["trim3"] is not None:
bowtie2_cmd += " --trim3 " + str(params["trim3"])
if "np" in params and params["np"] is not None:
bowtie2_cmd += " --np " + str(params["np"])
if "minins" in params and params["minins"] is not None:
bowtie2_cmd += " --minins " + str(params["minins"])
if "maxins" in params and params["maxins"] is not None:
bowtie2_cmd += " --maxins " + str(params["maxins"])
if "orientation" in params and params["orientation"] is not None:
bowtie2_cmd += " --" + params["orientation"]
if sample_type == "KBaseAssembly.SingleEndLibrary":
lib_type = "SingleEnd"
read_id = r_sample["data"]["handle"]["id"]
read_name = r_sample["data"]["handle"]["file_name"]
try:
script_util.download_file_from_shock(
logger,
shock_service_url=services["shock_service_url"],
shock_id=read_id,
filename=read_name,
directory=directory,
token=token,
)
bowtie2_cmd += " -U {0} -x {1} -S {2}".format(
os.path.join(directory, read_name), bowtie2_base, out_file
)
except Exception, e:
# logger.exception( "Unable to download shock file , {0}".format(read_name))
raise Exception("Unable to download shock file , {0}".format(read_name))
if sample_type == "KBaseAssembly.PairedEndLibrary":
lib_type = "PairedEnd"
read1_id = r_sample["data"]["handle_1"]["id"]
read1_name = r_sample["data"]["handle_1"]["file_name"]
read2_id = r_sample["data"]["handle_2"]["id"]
read2_name = r_sample["data"]["handle_2"]["file_name"]
try:
script_util.download_file_from_shock(
logger,
shock_service_url=services["shock_service_url"],
shock_id=read1_id,
filename=read1_name,
directory=directory,
token=token,
)
script_util.download_file_from_shock(
logger,
shock_service_url=services["shock_service_url"],
shock_id=read2_id,
filename=read2_name,
directory=directory,
token=token,
)
bowtie2_cmd += " -1 {0} -2 {1} -x {2} -S {3}".format(
os.path.join(directory, read1_name), os.path.join(directory, read2_name), bowtie2_base, out_file
)
except Exception, e:
# logger.Exception( "Unable to download shock file , {0} or {1}".format(read1_name,read2_name))
raise Exception("Unable to download shock file , {0} or {1}".format(read1_name, read2_name))
def runEach(self,task_params):
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']
read_sample = task_params['job_id']
condition = task_params['label']
directory = task_params['tophat_dir']
ws_id = task_params['ws_id']
reads_type = task_params['reads_type']
genome_id = task_params['annotation_id']
sampleset_id = task_params['sampleset_id']
gtf_file = task_params['gtf_file']
print "Downloading Read Sample{0}".format(read_sample)
logger.info("Downloading Read Sample{0}".format(read_sample))
try:
r_sample = ws_client.get_objects(
[{ 'name' : read_sample, 'workspace' : ws_id}])[0]
r_sample_info = ws_client.get_object_info_new({"objects": [{'name': read_sample, 'workspace': ws_id}]})[0]
sample_type = r_sample_info[2].split('-')[0]
output_name = read_sample.split('.')[0]+"_tophat_alignment"
output_dir = os.path.join(directory,output_name)
#if not os.path.exists(output_dir): os.makedirs(output_dir)
#out_file = output_dir +"/accepted_hits.sam"
bowtie2_base =os.path.join(directory,handler_util.get_file_with_suffix(directory,".rev.1.bt2"))
### Adding advanced options to Bowtie2Call
tophat_cmd = (' -p '+str(self.num_threads))
if('max_intron_length' in params and params['max_intron_length'] is not None ) : tophat_cmd += (' -I '+str(params['max_intron_length']))
if('min_intron_length' in params and params['min_intron_length'] is not None ): tophat_cmd += (' -i '+str(params['min_intron_length']))
if('min_anchor_length' in params and params['min_anchor_length'] is not None ): tophat_cmd += (' -a '+str(params['min_anchor_length']))
if('read_edit_dist' in params and params['read_edit_dist'] is not None ) : tophat_cmd += (' --read-edit-dist '+str(params['read_edit_dist']))
if('read_gap_length' in params and params['read_gap_length'] is not None) : tophat_cmd += (' --read-gap-length '+str(params['read_gap_length']))
if('read_mismatches' in params and params['read_mismatches'] is not None) : tophat_cmd += (' -N '+str(params['read_mismatches']))
if('library_type' in params and params['library_type'] is not None ) : tophat_cmd += (' --library-type ' + params['library_type'])
if('report_secondary_alignments' in params and int(params['report_secondary_alignments']) == 1) : tophat_cmd += ' --report-secondary-alignments'
if('no_coverage_search' in params and int(params['no_coverage_search']) == 1): tophat_cmd += ' --no-coverage-search'
if('preset_options' in params and params['preset_options'] is not None ): tophat_cmd += ' --'+params['preset_options']
#out_file = output_dir +"/accepted_hits.sam"
if sample_type == 'KBaseAssembly.SingleEndLibrary':
lib_type = 'SingleEnd'
read_id = r_sample['data']['handle']['id']
read_name = r_sample['data']['handle']['file_name']
try:
script_util.download_file_from_shock(self.logger, shock_service_url=self.urls['shock_service_url'], shock_id=read_id,filename=read_name, directory=directory,token=token)
tophat_cmd += ' -o {0} -G {1} {2} {3}'.format(output_dir,gtf_file,bowtie2_base,os.path.join(directory,read_name))
except Exception,e:
self.logger.exception(e)
raise Exception( "Unable to download shock file , {0}".format(read_name))
if sample_type == 'KBaseAssembly.PairedEndLibrary':
lib_type = 'PairedEnd'
if('orientation' in params and params['orientation'] is not None): tophat_cmd += ( ' --'+params['orientation'])
read1_id = r_sample['data']['handle_1']['id']
read1_name = r_sample['data']['handle_1']['file_name']
read2_id = r_sample['data']['handle_2']['id']
read2_name = r_sample['data']['handle_2']['file_name']
try:
script_util.download_file_from_shock(self.logger, shock_service_url=self.urls['shock_service_url'], shock_id=read1_id,filename=read1_name, directory=directory,token=token)
script_util.download_file_from_shock(self.logger, shock_service_url=self.urls['shock_service_url'], shock_id=read2_id,filename=read2_name, directory=directory,token=token)
tophat_cmd += ' -o {0} -G {1} {2} {3} {4}'.format(output_dir,gtf_file,bowtie2_base,os.path.join(directory,read1_name),os.path.join(directory,read2_name))
except Exception,e:
raise Exception( "Unable to download shock file , {0} or {1}".format(read1_name,read2_name))
def runEach(self,task_params):
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']
read_sample = task_params['job_id']
condition = task_params['label']
directory = task_params['bowtie2_dir']
ws_id = task_params['ws_id']
genome_id = task_params['annotation_id']
sampleset_id = task_params['sampleset_id']
print "Downloading Read Sample{0}".format(read_sample)
logger.info("Downloading Read Sample{0}".format(read_sample))
try:
#r_sample = ws_client.get_objects(
# [{ 'name' : read_sample, 'workspace' : ws_id}])[0]
r_sample = script_util.ws_get_obj(logger,ws_client, ws_id, read_sample)[0]
#r_sample_info = ws_client.get_object_info_new({"objects": [{'name': read_sample, 'workspace': ws_id}]})[0]
#sample_type = r_sample_info[2].split('-')[0]
sample_type = script_util.ws_get_type_name(logger, ws_client, ws_id, read_sample)
sample_name = script_util.ws_get_obj_name4file(self.logger, ws_client, ws_id, read_sample)
input_direc = os.path.join(directory,sample_name.split('.')[0]+"_bowtie2_input")
if not os.path.exists(input_direc): os.mkdir(input_direc)
output_name = sample_name.split('.')[0]+"_bowtie2_alignment"
output_dir = os.path.join(directory,output_name)
if not os.path.exists(output_dir): os.mkdir(output_dir)
base = handler_util.get_file_with_suffix(directory,".rev.1.bt2")
bowtie2_base =os.path.join(directory,base)
### Adding advanced options to Bowtie2Call
bowtie2_cmd = ''
bowtie2_cmd += ( ' -p {0}'.format(self.num_threads))
if('quality_score' in params and params['quality_score'] is not None): bowtie2_cmd += ( ' --'+params['quality_score'])
if('alignment_type' in params and params['alignment_type'] is not None): bowtie2_cmd += ( ' --'+params['alignment_type'] )
if('preset_options' in params and params['preset_options'] is not None ) and ('alignment_type' in params and params['alignment_type'] is not None):
if (params['alignment_type'] == 'local'):
bowtie2_cmd += (' --'+params['preset_options']+'-local')
else: bowtie2_cmd += (' --'+params['preset_options'] )
if('trim5' in params and params['trim5'] is not None): bowtie2_cmd += ( ' --trim5 '+str(params['trim5']))
if('trim3' in params and params['trim3'] is not None): bowtie2_cmd += ( ' --trim3 '+str(params['trim3']))
if('np' in params and params['np'] is not None): bowtie2_cmd += ( ' --np '+str(params['np']))
if('minins' in params and params['minins'] is not None): bowtie2_cmd += ( ' --minins '+str(params['minins']))
if('maxins' in params and params['maxins'] is not None): bowtie2_cmd += ( ' --maxins '+str(params['maxins']))
out_file = output_dir +"/accepted_hits.sam"
####
try:
sample_ref = script_util.ws_get_ref(self.logger, ws_client, ws_id, read_sample)
ds = script_util.ru_reads_download(self.logger, sample_ref,input_direc, token)
except Exception,e:
self.logger.exception(e)
raise Exception( "Unable to download reads file , {0}".format(read_sample))
if sample_type == 'KBaseAssembly.SingleEndLibrary' or sample_type == 'KBaseFile.SingleEndLibrary':
lib_type = 'SingleEnd'
bowtie2_cmd += " -U {0} -x {1} -S {2}".format(ds['fwd'],bowtie2_base,out_file)
if sample_type == 'KBaseAssembly.PairedEndLibrary' or sample_type == 'KBaseFile.PairedEndLibrary':
lib_type = 'PairedEnd'
if sample_type == 'KBaseAssembly.PairedEndLibrary':
if('orientation' in params and params['orientation'] is not None): hisat2_cmd += ( ' --'+params['orientation'])
else:
# TODO: the following can be read from PEL object
if('orientation' in params and params['orientation'] is not None): hisat2_cmd += ( ' --'+params['orientation'])
hisat2_cmd += " -1 {0} -2 {1} -x {2} -S {3}".format(ds['fwd'], ds['rev'],hisat2_base,out_file)
bowtie2_cmd += " -1 {0} -2 {1} -x {2} -S {3}".format(ds['fwd'], ds['rev'],bowtie2_base,out_file)
###
# if sample_type == 'KBaseAssembly.SingleEndLibrary' or sample_type == 'KBaseFile.SingleEndLibrary':
# lib_type = 'SingleEnd'
# if sample_type == 'KBaseAssembly.SingleEndLibrary':
# read_id = r_sample['data']['handle']['id']
# read_name = r_sample['data']['handle']['file_name']
# else:
# read_id = r_sample['data']['lib']['file']['id']
# read_name = r_sample['data']['lib']['file']['file_name']
# try:
# script_util.download_file_from_shock(self.logger, shock_service_url=self.urls['shock_service_url'], shock_id=read_id,filename=read_name, directory=input_direc,token=token)
# bowtie2_cmd += " -U {0} -x {1} -S {2}".format(os.path.join(input_direc,read_name),bowtie2_base,out_file)
# except Exception,e:
# self.logger.exception(e)
# raise Exception( "Unable to download shock file , {0}".format(read_name))
# if sample_type == 'KBaseAssembly.PairedEndLibrary' or sample_type == 'KBaseFile.PairedEndLibrary':
# lib_type = 'PairedEnd'
# if sample_type == 'KBaseAssembly.PairedEndLibrary':
# if('orientation' in params and params['orientation'] is not None): bowtie2_cmd += ( ' --'+params['orientation'])
# read1_id = r_sample['data']['handle_1']['id']
# read1_name = r_sample['data']['handle_1']['file_name']
# read2_id = r_sample['data']['handle_2']['id']
# read2_name = r_sample['data']['handle_2']['file_name']
# else:
# # TODO: the following can be read from PEL object
# if('orientation' in params and params['orientation'] is not None): bowtie2_cmd += ( ' --'+params['orientation'])
# read1_id = r_sample['data']['lib1']['file']['id']
# read1_name = r_sample['data']['lib1']['file']['file_name']
# read2_id = r_sample['data']['lib2']['file']['id']
# read2_name = r_sample['data']['lib2']['file']['file_name']
# try:
# script_util.download_file_from_shock(self.logger, shock_service_url=self.urls['shock_service_url'], shock_id=read1_id,filename=read1_name, directory=input_direc,token=token)
# script_util.download_file_from_shock(self.logger, shock_service_url=self.urls['shock_service_url'], shock_id=read2_id,filename=read2_name, directory=input_direc,token=token)
# bowtie2_cmd += " -1 {0} -2 {1} -x {2} -S {3}".format(os.path.join(input_direc,read1_name),os.path.join(input_direc,read2_name),bowtie2_base,out_file)
# except Exception,e:
# raise Exception( "Unable to download shock file , {0} or {1}".format(read1_name,read2_name))
try:
self.logger.info("Executing: bowtie2 {0}".format(bowtie2_cmd))
cmdline_output = script_util.runProgram(self.logger,"bowtie2",bowtie2_cmd,None,directory)
except Exception,e:
raise Exception("Failed to run command {0}".format(bowtie2_cmd))
def _CallTophat(
logger,
services,
ws_client,
hs,
ws_id,
sample_type,
num_threads,
read_sample,
gtf_file,
condition,
directory,
bowtie2index_id,
genome_id,
sampleset_id,
params,
token,
):
print "Downloading Read Sample{0}".format(read_sample)
if not logger:
logger = create_logger(directory, "run_Tophat_" + read_sample)
try:
r_sample = ws_client.get_objects([{"name": read_sample, "workspace": ws_id}])[0]
r_sample_info = ws_client.get_object_info_new({"objects": [{"name": read_sample, "workspace": ws_id}]})[0]
sample_type = r_sample_info[2].split("-")[0]
output_name = read_sample.split(".")[0] + "_tophat_alignment"
output_dir = os.path.join(directory, output_name)
# if not os.path.exists(output_dir): os.makedirs(output_dir)
# out_file = output_dir +"/accepted_hits.sam"
bowtie2_base = os.path.join(directory, handler_util.get_file_with_suffix(directory, ".rev.1.bt2"))
### Adding advanced options to tophat command
tophat_cmd = " -p " + str(num_threads)
if "max_intron_length" in params and params["max_intron_length"] is not None:
tophat_cmd += " -I " + str(params["max_intron_length"])
if "min_intron_length" in params and params["min_intron_length"] is not None:
tophat_cmd += " -i " + str(params["min_intron_length"])
if "min_anchor_length" in params and params["min_anchor_length"] is not None:
tophat_cmd += " -a " + str(params["min_anchor_length"])
if "read_edit_dist" in params and params["read_edit_dist"] is not None:
tophat_cmd += " --read-edit-dist " + str(params["read_edit_dist"])
if "read_gap_length" in params and params["read_gap_length"] is not None:
tophat_cmd += " --read-gap-length " + str(params["read_gap_length"])
if "read_mismatches" in params and params["read_mismatches"] is not None:
tophat_cmd += " -N " + str(params["read_mismatches"])
if "library_type" in params and params["library_type"] is not None:
tophat_cmd += " --library-type " + params["library_type"]
if "report_secondary_alignments" in params and int(params["report_secondary_alignments"]) == 1:
tophat_cmd += " --report-secondary-alignments"
if "no_coverage_search" in params and int(params["no_coverage_search"]) == 1:
tophat_cmd += " --no-coverage-search"
if "preset_options" in params and params["preset_options"] is not None:
tophat_cmd += " --" + params["preset_options"]
if sample_type == "KBaseAssembly.SingleEndLibrary":
lib_type = "SingleEnd"
read_id = r_sample["data"]["handle"]["id"]
read_name = r_sample["data"]["handle"]["file_name"]
try:
script_util.download_file_from_shock(
logger,
shock_service_url=services["shock_service_url"],
shock_id=read_id,
filename=read_name,
directory=directory,
token=token,
)
tophat_cmd += " -o {0} -G {1} {2} {3}".format(
output_dir, gtf_file, bowtie2_base, os.path.join(directory, read_name)
)
except Exception, e:
raise Exception("Unable to download shock file , {0}".format(read_name))
if sample_type == "KBaseAssembly.PairedEndLibrary":
lib_type = "PairedEnd"
read1_id = r_sample["data"]["handle_1"]["id"]
read1_name = r_sample["data"]["handle_1"]["file_name"]
read2_id = r_sample["data"]["handle_2"]["id"]
read2_name = r_sample["data"]["handle_2"]["file_name"]
try:
script_util.download_file_from_shock(
logger,
shock_service_url=services["shock_service_url"],
shock_id=read1_id,
filename=read1_name,
directory=directory,
token=token,
)
script_util.download_file_from_shock(
logger,
shock_service_url=services["shock_service_url"],
shock_id=read2_id,
filename=read2_name,
directory=directory,
token=token,
)
tophat_cmd += " -o {0} -G {1} {2} {3} {4}".format(
output_dir,
gtf_file,
bowtie2_base,
os.path.join(directory, read1_name),
os.path.join(directory, read2_name),
)
except Exception, e:
raise Exception("Unable to download shock file , {0} or {1}".format(read1_name, read2_name))
def runEach(self, task_params):
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']
read_sample = task_params['job_id']
condition = task_params['label']
directory = task_params['hisat2_dir']
ws_id = task_params['ws_id']
genome_id = task_params['annotation_id']
sampleset_id = task_params['sampleset_id']
print "Downloading Read Sample{0}".format(read_sample)
logger.info("Downloading Read Sample{0}".format(read_sample))
try:
#r_sample = ws_client.get_objects(
# [{ 'name' : read_sample, 'workspace' : ws_id}])[0]
#r_sample_info = ws_client.get_object_info_new({"objects": [{'name': read_sample, 'workspace': ws_id}]})[0]
#sample_type = r_sample_info[2].split('-')[0]
r_sample = script_util.ws_get_obj(self.logger, ws_client, ws_id,
read_sample)[0]
sample_type = script_util.ws_get_type_name(self.logger, ws_client,
ws_id, read_sample)
sample_name = script_util.ws_get_obj_name4file(
self.logger, ws_client, ws_id, read_sample)
input_direc = os.path.join(
directory,
sample_name.split('.')[0] + "_hisat2_input")
if not os.path.exists(input_direc): os.mkdir(input_direc)
output_name = sample_name.split('.')[0] + "_hisat2_alignment"
output_dir = os.path.join(directory, output_name)
if not os.path.exists(output_dir): os.mkdir(output_dir)
print directory
base = handler_util.get_file_with_suffix(directory, ".1.ht2")
print base
hisat2_base = os.path.join(directory, base)
### Adding advanced options to Bowtie2Call
hisat2_cmd = ''
hisat2_cmd += (' -p {0}'.format(self.num_threads))
if ('quality_score' in params
and params['quality_score'] is not None):
hisat2_cmd += (' --' + params['quality_score'])
if ('alignment_type' in params
and params['alignment_type'] is not None):
hisat2_cmd += (' --' + params['alignment_type'])
if ('trim5' in params and params['trim5'] is not None):
hisat2_cmd += (' --trim5 ' + str(params['trim5']))
if ('trim3' in params and params['trim3'] is not None):
hisat2_cmd += (' --trim3 ' + str(params['trim3']))
if ('np' in params and params['np'] is not None):
hisat2_cmd += (' --np ' + str(params['np']))
if ('minins' in params and params['minins'] is not None):
hisat2_cmd += (' --minins ' + str(params['minins']))
if ('maxins' in params and params['maxins'] is not None):
hisat2_cmd += (' --maxins ' + str(params['maxins']))
#if('orientation' in params and params['orientation'] is not None): hisat2_cmd += ( ' --'+params['orientation'])
if ('min_intron_length' in params
and params['min_intron_length'] is not None):
hisat2_cmd += (' --min-intronlen ' +
str(params['min_intron_length']))
if ('max_intron_length' in params
and params['max_intron_length'] is not None):
hisat2_cmd += (' --max-intronlen ' +
str(params['max_intron_length']))
if ('no_spliced_alignment' in params
and params['no_spliced_alignment'] != 0):
hisat2_cmd += (' --no-spliced-alignment')
if ('transcriptome_mapping_only' in params
and params['transcriptome_mapping_only'] != 0):
hisat2_cmd += (' --transcriptome-mapping-only')
if ('tailor_alignments' in params
and params['tailor_alignments'] is not None):
hisat2_cmd += (' --' + params['tailor_alignments'])
out_file = output_dir + "/accepted_hits.sam"
####
try:
sample_ref = script_util.ws_get_ref(self.logger, ws_client,
ws_id, read_sample)
ds = script_util.ru_reads_download(self.logger, sample_ref,
input_direc, token)
self.logger.info(ds)
except Exception, e:
self.logger.exception(e)
raise Exception(
"Unable to download reads file , {0}".format(read_sample))
if sample_type == 'KBaseAssembly.SingleEndLibrary' or sample_type == 'KBaseFile.SingleEndLibrary':
lib_type = 'SingleEnd'
hisat2_cmd += " -U {0} -x {1} -S {2}".format(
ds['fwd'], hisat2_base, out_file)
if sample_type == 'KBaseAssembly.PairedEndLibrary' or sample_type == 'KBaseFile.PairedEndLibrary':
lib_type = 'PairedEnd'
if sample_type == 'KBaseAssembly.PairedEndLibrary':
if ('orientation' in params
and params['orientation'] is not None):
hisat2_cmd += (' --' + params['orientation'])
else:
# TODO: the following can be read from PEL object
if ('orientation' in params
and params['orientation'] is not None):
hisat2_cmd += (' --' + params['orientation'])
hisat2_cmd += " -1 {0} -2 {1} -x {2} -S {3}".format(
ds['fwd'], ds['rev'], hisat2_base, out_file)
#if sample_type == 'KBaseAssembly.SingleEndLibrary' or sample_type == 'KBaseFile.SingleEndLibrary':
# lib_type = 'SingleEnd'
# if sample_type == 'KBaseAssembly.SingleEndLibrary':
# read_id = r_sample['data']['handle']['id']
# read_name = r_sample['data']['handle']['file_name']
# else:
# read_id = r_sample['data']['lib']['file']['id']
# read_name = r_sample['data']['lib']['file']['file_name']
# try:
# script_util.download_file_from_shock(self.logger, shock_service_url=self.urls['shock_service_url'], shock_id=read_id,filename=read_name, directory=input_direc,token=token)
# hisat2_cmd += " -U {0} -x {1} -S {2}".format(os.path.join(input_direc,read_name),hisat2_base,out_file)
# except Exception,e:
# self.logger.exception(e)
# raise Exception( "Unable to download shock file , {0}".format(read_name))
#if sample_type == 'KBaseAssembly.PairedEndLibrary' or sample_type == 'KBaseFile.PairedEndLibrary':
# lib_type = 'PairedEnd'
# if sample_type == 'KBaseAssembly.PairedEndLibrary':
# if('orientation' in params and params['orientation'] is not None): hisat2_cmd += ( ' --'+params['orientation'])
# read1_id = r_sample['data']['handle_1']['id']
# read1_name = r_sample['data']['handle_1']['file_name']
# read2_id = r_sample['data']['handle_2']['id']
# read2_name = r_sample['data']['handle_2']['file_name']
# else:
# # TODO: the following can be read from PEL object
# if('orientation' in params and params['orientation'] is not None): hisat2_cmd += ( ' --'+params['orientation'])
# read1_id = r_sample['data']['lib1']['file']['id']
# read1_name = r_sample['data']['lib1']['file']['file_name']
# read2_id = r_sample['data']['lib2']['file']['id']
# read2_name = r_sample['data']['lib2']['file']['file_name']
# try:
# script_util.download_file_from_shock(self.logger, shock_service_url=self.urls['shock_service_url'], shock_id=read1_id,filename=read1_name, directory=input_direc,token=token)
# script_util.download_file_from_shock(self.logger, shock_service_url=self.urls['shock_service_url'], shock_id=read2_id,filename=read2_name, directory=input_direc,token=token)
# hisat2_cmd += " -1 {0} -2 {1} -x {2} -S {3}".format(os.path.join(input_direc,read1_name),os.path.join(input_direc,read2_name),hisat2_base,out_file)
# except Exception,e:
# logger.exception(e)
# raise Exception( "Unable to download shock file , {0} or {1}".format(read1_name,read2_name))
try:
self.logger.info("Executing: hisat2 {0}".format(hisat2_cmd))
cmdline_output = script_util.runProgram(
self.logger, "hisat2", hisat2_cmd, None, directory)
except Exception, e:
logger.exception(e)
raise Exception("Failed to run command {0}".format(hisat2_cmd))
def _CallBowtie2(logger, services, ws_client, hs, ws_id, sample_type,
num_threads, read_sample, condition, directory,
bowtie2index_id, genome_id, sampleset_id, params, token):
#logger.info("Downloading Read Sample{0}".format(read_sample))
print "Downloading Read Sample{0}".format(read_sample)
if not logger:
logger = create_logger(directory, "run_Bowtie2_" + read_sample)
logger.info("Downloading Read Sample{0}".format(read_sample))
try:
r_sample = ws_client.get_objects([{
'name': read_sample,
'workspace': ws_id
}])[0]
r_sample_info = ws_client.get_object_info_new(
{"objects": [{
'name': read_sample,
'workspace': ws_id
}]})[0]
sample_type = r_sample_info[2].split('-')[0]
output_name = read_sample.split('.')[0] + "_bowtie2_alignment"
output_dir = os.path.join(directory, output_name)
if not os.path.exists(output_dir): os.mkdir(output_dir)
out_file = output_dir + "/accepted_hits.sam"
bowtie2_base = os.path.join(
directory,
handler_util.get_file_with_suffix(directory, ".rev.1.bt2"))
### Adding advanced options to Bowtie2Call
bowtie2_cmd = ''
bowtie2_cmd += (' -p {0}'.format(num_threads))
if ('quality_score' in params and params['quality_score'] is not None):
bowtie2_cmd += (' --' + params['quality_score'])
if ('alignment_type' in params
and params['alignment_type'] is not None):
bowtie2_cmd += (' --' + params['alignment_type'])
if ('preset_options' in params and params['preset_options']
is not None) and ('alignment_type' in params
and params['alignment_type'] is not None):
if (params['alignment_type'] == 'local'):
bowtie2_cmd += (' --' + params['preset_options'] + '-local')
else:
bowtie2_cmd += (' --' + params['preset_options'])
if ('trim5' in params and params['trim5'] is not None):
bowtie2_cmd += (' --trim5 ' + str(params['trim5']))
if ('trim3' in params and params['trim3'] is not None):
bowtie2_cmd += (' --trim3 ' + str(params['trim3']))
if ('np' in params and params['np'] is not None):
bowtie2_cmd += (' --np ' + str(params['np']))
if ('minins' in params and params['minins'] is not None):
bowtie2_cmd += (' --minins ' + str(params['minins']))
if ('maxins' in params and params['maxins'] is not None):
bowtie2_cmd += (' --maxins ' + str(params['maxins']))
if ('orientation' in params and params['orientation'] is not None):
bowtie2_cmd += (' --' + params['orientation'])
if sample_type == 'KBaseAssembly.SingleEndLibrary':
lib_type = 'SingleEnd'
read_id = r_sample['data']['handle']['id']
read_name = r_sample['data']['handle']['file_name']
try:
script_util.download_file_from_shock(
logger,
shock_service_url=services['shock_service_url'],
shock_id=read_id,
filename=read_name,
directory=directory,
token=token)
bowtie2_cmd += " -U {0} -x {1} -S {2}".format(
os.path.join(directory, read_name), bowtie2_base, out_file)
except Exception, e:
#logger.exception( "Unable to download shock file , {0}".format(read_name))
raise Exception(
"Unable to download shock file , {0}".format(read_name))
if sample_type == 'KBaseAssembly.PairedEndLibrary':
lib_type = 'PairedEnd'
read1_id = r_sample['data']['handle_1']['id']
read1_name = r_sample['data']['handle_1']['file_name']
read2_id = r_sample['data']['handle_2']['id']
read2_name = r_sample['data']['handle_2']['file_name']
try:
script_util.download_file_from_shock(
logger,
shock_service_url=services['shock_service_url'],
shock_id=read1_id,
filename=read1_name,
directory=directory,
token=token)
script_util.download_file_from_shock(
logger,
shock_service_url=services['shock_service_url'],
shock_id=read2_id,
filename=read2_name,
directory=directory,
token=token)
bowtie2_cmd += " -1 {0} -2 {1} -x {2} -S {3}".format(
os.path.join(directory, read1_name),
os.path.join(directory, read2_name), bowtie2_base,
out_file)
except Exception, e:
#logger.Exception( "Unable to download shock file , {0} or {1}".format(read1_name,read2_name))
raise Exception(
"Unable to download shock file , {0} or {1}".format(
read1_name, read2_name))
def runEach(self,task_params):
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']
read_sample = task_params['job_id']
condition = task_params['label']
directory = task_params['tophat_dir']
ws_id = task_params['ws_id']
genome_id = task_params['annotation_id']
sampleset_id = task_params['sampleset_id']
gtf_file = task_params['gtf_file']
print "Downloading Read Sample{0}".format(read_sample)
logger.info("Downloading Read Sample{0}".format(read_sample))
try:
#r_sample = ws_client.get_objects(
# [{ 'name' : read_sample, 'workspace' : ws_id}])[0]
r_sample = script_util.ws_get_obj(logger,ws_client, ws_id, read_sample)[0]
#r_sample_info = ws_client.get_object_info_new({"objects": [{'name': read_sample, 'workspace': ws_id}]})[0]
#sample_type = r_sample_info[2].split('-')[0]
sample_type = script_util.ws_get_type_name(logger, ws_client, ws_id, read_sample)
sample_name = script_util.ws_get_obj_name4file(self.logger, ws_client, ws_id, read_sample)
output_name = sample_name.split('.')[0]+"_tophat_alignment"
output_dir = os.path.join(directory,output_name)
#if not os.path.exists(output_dir): os.makedirs(output_dir)
#out_file = output_dir +"/accepted_hits.sam"
bowtie2_base =os.path.join(directory,handler_util.get_file_with_suffix(directory,".rev.1.bt2"))
### Adding advanced options to Bowtie2Call
tophat_cmd = (' -p '+str(self.num_threads))
if('max_intron_length' in params and params['max_intron_length'] is not None ) : tophat_cmd += (' -I '+str(params['max_intron_length']))
if('min_intron_length' in params and params['min_intron_length'] is not None ): tophat_cmd += (' -i '+str(params['min_intron_length']))
if('min_anchor_length' in params and params['min_anchor_length'] is not None ): tophat_cmd += (' -a '+str(params['min_anchor_length']))
if('read_edit_dist' in params and params['read_edit_dist'] is not None ) : tophat_cmd += (' --read-edit-dist '+str(params['read_edit_dist']))
if('read_gap_length' in params and params['read_gap_length'] is not None) : tophat_cmd += (' --read-gap-length '+str(params['read_gap_length']))
if('read_mismatches' in params and params['read_mismatches'] is not None) : tophat_cmd += (' -N '+str(params['read_mismatches']))
if('library_type' in params and params['library_type'] is not None ) : tophat_cmd += (' --library-type ' + params['library_type'])
if('report_secondary_alignments' in params and int(params['report_secondary_alignments']) == 1) : tophat_cmd += ' --report-secondary-alignments'
if('no_coverage_search' in params and int(params['no_coverage_search']) == 1): tophat_cmd += ' --no-coverage-search'
if('preset_options' in params and params['preset_options'] is not None ): tophat_cmd += ' --'+params['preset_options']
#out_file = output_dir +"/accepted_hits.sam"
try:
sample_ref = script_util.ws_get_ref(self.logger, ws_client, ws_id, read_sample)
ds = script_util.ru_reads_download(self.logger, sample_ref,directory, token)
except Exception,e:
self.logger.exception(e)
raise Exception( "Unable to download reads file , {0}".format(read_sample))
if sample_type == 'KBaseAssembly.SingleEndLibrary' or sample_type == 'KBaseFile.SingleEndLibrary':
lib_type = 'SingleEnd'
tophat_cmd += ' -o {0} -G {1} {2} {3}'.format(output_dir,gtf_file,bowtie2_base,ds['fwd'])
if sample_type == 'KBaseAssembly.PairedEndLibrary' or sample_type == 'KBaseFile.PairedEndLibrary':
lib_type = 'PairedEnd'
if sample_type == 'KBaseAssembly.PairedEndLibrary':
if('orientation' in params and params['orientation'] is not None): tophat_cmd += ( ' --'+params['orientation'])
else:
# TODO: the following can be read from PEL object
if('orientation' in params and params['orientation'] is not None): tophat_cmd += ( ' --'+params['orientation'])
tophat_cmd += ' -o {0} -G {1} {2} {3} {4}'.format(output_dir,gtf_file,bowtie2_base,ds['fwd'],ds['rev'])
# if sample_type == 'KBaseAssembly.SingleEndLibrary' or sample_type == 'KBaseFile.SingleEndLibrary':
# lib_type = 'SingleEnd'
# if sample_type == 'KBaseAssembly.SingleEndLibrary':
# read_id = r_sample['data']['handle']['id']
# read_name = r_sample['data']['handle']['file_name']
# else:
# read_id = r_sample['data']['lib']['file']['id']
# read_name = r_sample['data']['lib']['file']['file_name']
# try:
# script_util.download_file_from_shock(self.logger, shock_service_url=self.urls['shock_service_url'], shock_id=read_id,filename=read_name, directory=directory,token=token)
# tophat_cmd += ' -o {0} -G {1} {2} {3}'.format(output_dir,gtf_file,bowtie2_base,os.path.join(directory,read_name))
# except Exception,e:
# self.logger.exception(e)
# raise Exception( "Unable to download shock file , {0}".format(read_name))
# if sample_type == 'KBaseAssembly.PairedEndLibrary' or sample_type == 'KBaseFile.PairedEndLibrary':
# lib_type = 'PairedEnd'
# if sample_type == 'KBaseAssembly.PairedEndLibrary':
# if('orientation' in params and params['orientation'] is not None): tophat_cmd += ( ' --'+params['orientation'])
# read1_id = r_sample['data']['handle_1']['id']
# read1_name = r_sample['data']['handle_1']['file_name']
# read2_id = r_sample['data']['handle_2']['id']
# read2_name = r_sample['data']['handle_2']['file_name']
# else:
# # TODO: the following can be read from PEL object
# if('orientation' in params and params['orientation'] is not None): tophat_cmd += ( ' --'+params['orientation'])
# read1_id = r_sample['data']['lib1']['file']['id']
# read1_name = r_sample['data']['lib1']['file']['file_name']
# read2_id = r_sample['data']['lib2']['file']['id']
# read2_name = r_sample['data']['lib2']['file']['file_name']
# try:
# script_util.download_file_from_shock(self.logger, shock_service_url=self.urls['shock_service_url'], shock_id=read1_id,filename=read1_name, directory=directory,token=token)
# script_util.download_file_from_shock(self.logger, shock_service_url=self.urls['shock_service_url'], shock_id=read2_id,filename=read2_name, directory=directory,token=token)
# tophat_cmd += ' -o {0} -G {1} {2} {3} {4}'.format(output_dir,gtf_file,bowtie2_base,os.path.join(directory,read1_name),os.path.join(directory,read2_name))
# except Exception,e:
# raise Exception( "Unable to download shock file , {0} or {1}".format(read1_name,read2_name))
try:
self.logger.info("Executing: tophat {0}".format(tophat_cmd))
cmdline_output, cmd_err = script_util.runProgram(self.logger,"tophat",tophat_cmd,None,directory)
except Exception,e:
raise Exception("Failed to run command {0}\n{1}\n{2}".format(tophat_cmd,cmdline_output,cmd_err))
def runEach(self,task_params):
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']
read_sample = task_params['job_id']
condition = task_params['label']
directory = task_params['bowtie2_dir']
ws_id = task_params['ws_id']
reads_type = task_params['reads_type']
genome_id = task_params['annotation_id']
sampleset_id = task_params['sampleset_id']
print "Downloading Read Sample{0}".format(read_sample)
logger.info("Downloading Read Sample{0}".format(read_sample))
try:
r_sample = ws_client.get_objects(
[{ 'name' : read_sample, 'workspace' : ws_id}])[0]
r_sample_info = ws_client.get_object_info_new({"objects": [{'name': read_sample, 'workspace': ws_id}]})[0]
sample_type = r_sample_info[2].split('-')[0]
input_direc = os.path.join(directory,read_sample.split('.')[0]+"_bowtie2_input")
if not os.path.exists(input_direc): os.mkdir(input_direc)
output_name = read_sample.split('.')[0]+"_bowtie2_alignment"
output_dir = os.path.join(directory,output_name)
if not os.path.exists(output_dir): os.mkdir(output_dir)
base = handler_util.get_file_with_suffix(directory,".rev.1.bt2")
bowtie2_base =os.path.join(directory,base)
### Adding advanced options to Bowtie2Call
bowtie2_cmd = ''
bowtie2_cmd += ( ' -p {0}'.format(self.num_threads))
if('quality_score' in params and params['quality_score'] is not None): bowtie2_cmd += ( ' --'+params['quality_score'])
if('alignment_type' in params and params['alignment_type'] is not None): bowtie2_cmd += ( ' --'+params['alignment_type'] )
if('preset_options' in params and params['preset_options'] is not None ) and ('alignment_type' in params and params['alignment_type'] is not None):
if (params['alignment_type'] == 'local'):
bowtie2_cmd += (' --'+params['preset_options']+'-local')
else: bowtie2_cmd += (' --'+params['preset_options'] )
if('trim5' in params and params['trim5'] is not None): bowtie2_cmd += ( ' --trim5 '+str(params['trim5']))
if('trim3' in params and params['trim3'] is not None): bowtie2_cmd += ( ' --trim3 '+str(params['trim3']))
if('np' in params and params['np'] is not None): bowtie2_cmd += ( ' --np '+str(params['np']))
if('minins' in params and params['minins'] is not None): bowtie2_cmd += ( ' --minins '+str(params['minins']))
if('maxins' in params and params['maxins'] is not None): bowtie2_cmd += ( ' --maxins '+str(params['maxins']))
out_file = output_dir +"/accepted_hits.sam"
if sample_type == 'KBaseAssembly.SingleEndLibrary':
lib_type = 'SingleEnd'
read_id = r_sample['data']['handle']['id']
read_name = r_sample['data']['handle']['file_name']
try:
script_util.download_file_from_shock(self.logger, shock_service_url=self.urls['shock_service_url'], shock_id=read_id,filename=read_name, directory=input_direc,token=token)
bowtie2_cmd += " -U {0} -x {1} -S {2}".format(os.path.join(input_direc,read_name),bowtie2_base,out_file)
except Exception,e:
self.logger.exception(e)
raise Exception( "Unable to download shock file , {0}".format(read_name))
if sample_type == 'KBaseAssembly.PairedEndLibrary':
lib_type = 'PairedEnd'
if('orientation' in params and params['orientation'] is not None): bowtie2_cmd += ( ' --'+params['orientation'])
read1_id = r_sample['data']['handle_1']['id']
read1_name = r_sample['data']['handle_1']['file_name']
read2_id = r_sample['data']['handle_2']['id']
read2_name = r_sample['data']['handle_2']['file_name']
try:
script_util.download_file_from_shock(self.logger, shock_service_url=self.urls['shock_service_url'], shock_id=read1_id,filename=read1_name, directory=input_direc,token=token)
script_util.download_file_from_shock(self.logger, shock_service_url=self.urls['shock_service_url'], shock_id=read2_id,filename=read2_name, directory=input_direc,token=token)
bowtie2_cmd += " -1 {0} -2 {1} -x {2} -S {3}".format(os.path.join(input_direc,read1_name),os.path.join(output_dir,read2_name),bowtie2_base,out_file)
except Exception,e:
raise Exception( "Unable to download shock file , {0} or {1}".format(read1_name,read2_name))
reads = sample['data']['sample_ids']
reads_type= sample['data']['Library_type']
if reads_type == 'PairedEnd': r_type = 'KBaseAssembly.PairedEndLibrary'
else: r_type = 'KBaseAssembly.SingleEndLibrary'
e_ws_objs = script_util.if_ws_obj_exists(None,ws_client,params['ws_id'],r_type,reads)
missing_objs = [i for i in reads if not i in e_ws_objs]
if len(e_ws_objs) != len(reads):
raise Exception('Missing Library objects {0} in the {1}. please copy them and run this method'.format(",".join(missing_objs),params['ws_id']))
### Build Hisat2 index
fasta_file = script_util.generate_fasta(logger,services,token,annotation_id,hisat2_dir,params['genome_id'])
logger.info("Sanitizing the fasta file to correct id names {}".format(datetime.datetime.utcnow()))
mapping_filename = c_mapping.create_sanitized_contig_ids(fasta_file)
c_mapping.replace_fasta_contig_ids(fasta_file, mapping_filename, to_modified=True)
logger.info("Generating FASTA file completed successfully : {}".format(datetime.datetime.utcnow()))
hisat2base =os.path.join(hisat2_dir,handler_util.get_file_with_suffix(hisat2_dir,".fasta"))
hisat2base_cmd = '{0} {1}'.format(fasta_file,hisat2base)
try:
logger.info("Building Index for Hisat2 {0}".format(hisat2base_cmd))
cmdline_output = script_util.runProgram(logger,"hisat2-build",hisat2base_cmd,None,hisat2_dir)
except Exception,e:
raise Exception("Failed to run command {0}".format(hisat2base_cmd))
ws_gtf = params['genome_id']+"_GTF"
ret = script_util.if_obj_exists(None,ws_client,params['ws_id'],"KBaseRNASeq.GFFAnnotation",[ws_gtf])
print ret
if not ret is None:
logger.info("GFF Annotation Exist for Genome Annotation {0}.... Skipping step ".format(params['genome_id']))
annot_name,annot_id = ret[0]
gtf_obj=ws_client.get_objects([{'ref' : annot_id}])[0]
gtf_id=gtf_obj['data']['handle']['id']
gtf_name=gtf_obj['data']['handle']['file_name']
bw_name = bowtie_index['data']['handle']['file_name']
genome_id = bowtie_index['data']['genome_id']
annotation_gtf = ws_client.get_object_info([{"ref" :genome_id}],includeMetadata=None)[0][1]
shared_files={}
shared_files[bw_name] = bw_id
script_util.download_shock_files(logger,self.urls['shock_service_url'],tophat_dir,shared_files,token)
try:
logger.info("Unzipping Bowtie2 Indices")
script_util.unzip_files(logger,os.path.join(tophat_dir,bw_name),tophat_dir)
mv_dir= handler_util.get_dir(tophat_dir)
if mv_dir is not None:
script_util.move_files(logger,mv_dir,tophat_dir)
except Exception, e:
logger.error("".join(traceback.format_exc()))
raise Exception("Unzip indexfile error")
fasta_file =os.path.join(tophat_dir,(handler_util.get_file_with_suffix(tophat_dir,".fa")+".fa"))
bowtie2base =os.path.join(tophat_dir,handler_util.get_file_with_suffix(tophat_dir,".rev.1.bt2"))
### Check if GTF annotation object exist or skip this step
### Check if the gtf object exists in the workspace
### Only run create_gtf_annotation if object doesnt exist
ws_gtf = annotation_gtf+"_GTF_Annotation"
genome_name = script_util.ws_get_obj_name( logger, ws_client, params['ws_id'], genome_id )
gtf_file = script_util.check_and_download_existing_handle_obj(logger,ws_client,self.urls,params['ws_id'],ws_gtf,"KBaseRNASeq.GFFAnnotation",tophat_dir,token)
if gtf_file is None:
gtf_file = rnaseq_util.create_gtf_annotation_from_genome(logger,ws_client,hs,self.urls,params['ws_id'],genome_id,genome_name,tophat_dir,token)
#ret = script_util.if_obj_exists(None,ws_client,params['ws_id'],"KBaseRNASeq.GFFAnnotation",[ws_gtf]) # this line should be safe from reference
#if not ret is None:
# logger.info("GFF Annotation Exist for Genome Annotation {0}.... Skipping step ".format(annotation_gtf))
# annot_name,annot_id = ret[0]
script_util.download_shock_files(logger,
self.urls['shock_service_url'],
tophat_dir, shared_files, token)
try:
logger.info("Unzipping Bowtie2 Indices")
script_util.unzip_files(logger, os.path.join(tophat_dir, bw_name),
tophat_dir)
mv_dir = handler_util.get_dir(tophat_dir)
if mv_dir is not None:
script_util.move_files(logger, mv_dir, tophat_dir)
except Exception, e:
logger.error("".join(traceback.format_exc()))
raise Exception("Unzip indexfile error")
fasta_file = os.path.join(
tophat_dir,
(handler_util.get_file_with_suffix(tophat_dir, ".fa") + ".fa"))
bowtie2base = os.path.join(
tophat_dir,
handler_util.get_file_with_suffix(tophat_dir, ".rev.1.bt2"))
### Check if GTF annotation object exist or skip this step
### Check if the gtf object exists in the workspace
### Only run create_gtf_annotation if object doesnt exist
ws_gtf = annotation_gtf + "_GTF_Annotation"
gtf_file = script_util.check_and_download_existing_handle_obj(
logger, ws_client, self.urls, params['ws_id'], ws_gtf,
"KBaseRNASeq.GFFAnnotation", tophat_dir, token)
if gtf_file is None:
gtf_file = rnaseq_util.create_gtf_annotation_from_genome(
logger, ws_client, hs, self.urls, params['ws_id'], ref_id,
genome_name, tophat_dir, token)
def _CallTophat(logger, services, ws_client, hs, ws_id, sample_type,
num_threads, read_sample, gtf_file, condition, directory,
bowtie2index_id, genome_id, sampleset_id, params, token):
print "Downloading Read Sample{0}".format(read_sample)
if not logger:
logger = create_logger(directory, "run_Tophat_" + read_sample)
try:
r_sample = ws_client.get_objects([{
'name': read_sample,
'workspace': ws_id
}])[0]
r_sample_info = ws_client.get_object_info_new(
{"objects": [{
'name': read_sample,
'workspace': ws_id
}]})[0]
sample_type = r_sample_info[2].split('-')[0]
output_name = read_sample.split('.')[0] + "_tophat_alignment"
output_dir = os.path.join(directory, output_name)
#if not os.path.exists(output_dir): os.makedirs(output_dir)
#out_file = output_dir +"/accepted_hits.sam"
bowtie2_base = os.path.join(
directory,
handler_util.get_file_with_suffix(directory, ".rev.1.bt2"))
### Adding advanced options to tophat command
tophat_cmd = (' -p ' + str(num_threads))
if ('max_intron_length' in params
and params['max_intron_length'] is not None):
tophat_cmd += (' -I ' + str(params['max_intron_length']))
if ('min_intron_length' in params
and params['min_intron_length'] is not None):
tophat_cmd += (' -i ' + str(params['min_intron_length']))
if ('min_anchor_length' in params
and params['min_anchor_length'] is not None):
tophat_cmd += (' -a ' + str(params['min_anchor_length']))
if ('read_edit_dist' in params
and params['read_edit_dist'] is not None):
tophat_cmd += (' --read-edit-dist ' +
str(params['read_edit_dist']))
if ('read_gap_length' in params
and params['read_gap_length'] is not None):
tophat_cmd += (' --read-gap-length ' +
str(params['read_gap_length']))
if ('read_mismatches' in params
and params['read_mismatches'] is not None):
tophat_cmd += (' -N ' + str(params['read_mismatches']))
if ('library_type' in params and params['library_type'] is not None):
tophat_cmd += (' --library-type ' + params['library_type'])
if ('report_secondary_alignments' in params
and int(params['report_secondary_alignments']) == 1):
tophat_cmd += ' --report-secondary-alignments'
if ('no_coverage_search' in params
and int(params['no_coverage_search']) == 1):
tophat_cmd += ' --no-coverage-search'
if ('preset_options' in params
and params['preset_options'] is not None):
tophat_cmd += ' --' + params['preset_options']
if sample_type == 'KBaseAssembly.SingleEndLibrary':
lib_type = 'SingleEnd'
read_id = r_sample['data']['handle']['id']
read_name = r_sample['data']['handle']['file_name']
try:
script_util.download_file_from_shock(
logger,
shock_service_url=services['shock_service_url'],
shock_id=read_id,
filename=read_name,
directory=directory,
token=token)
tophat_cmd += ' -o {0} -G {1} {2} {3}'.format(
output_dir, gtf_file, bowtie2_base,
os.path.join(directory, read_name))
except Exception, e:
raise Exception(
"Unable to download shock file , {0}".format(read_name))
if sample_type == 'KBaseAssembly.PairedEndLibrary':
lib_type = 'PairedEnd'
read1_id = r_sample['data']['handle_1']['id']
read1_name = r_sample['data']['handle_1']['file_name']
read2_id = r_sample['data']['handle_2']['id']
read2_name = r_sample['data']['handle_2']['file_name']
try:
script_util.download_file_from_shock(
logger,
shock_service_url=services['shock_service_url'],
shock_id=read1_id,
filename=read1_name,
directory=directory,
token=token)
script_util.download_file_from_shock(
logger,
shock_service_url=services['shock_service_url'],
shock_id=read2_id,
filename=read2_name,
directory=directory,
token=token)
tophat_cmd += ' -o {0} -G {1} {2} {3} {4}'.format(
output_dir, gtf_file, bowtie2_base,
os.path.join(directory, read1_name),
os.path.join(directory, read2_name))
except Exception, e:
raise Exception(
"Unable to download shock file , {0} or {1}".format(
read1_name, read2_name))
bw_name = bowtie_index['data']['handle']['file_name']
genome_id = bowtie_index['data']['genome_id']
annotation_gtf = ws_client.get_object_info([{"ref" :genome_id}],includeMetadata=None)[0][1]
shared_files={}
shared_files[bw_name] = bw_id
script_util.download_shock_files(logger,self.urls['shock_service_url'],tophat_dir,shared_files,token)
try:
logger.info("Unzipping Bowtie2 Indices")
script_util.unzip_files(logger,os.path.join(tophat_dir,bw_name),tophat_dir)
mv_dir= handler_util.get_dir(tophat_dir)
if mv_dir is not None:
script_util.move_files(logger,mv_dir,tophat_dir)
except Exception, e:
logger.error("".join(traceback.format_exc()))
raise Exception("Unzip indexfile error: Please contact [email protected]")
fasta_file =os.path.join(tophat_dir,(handler_util.get_file_with_suffix(tophat_dir,".fa")+".fa"))
bowtie2base =os.path.join(tophat_dir,handler_util.get_file_with_suffix(tophat_dir,".rev.1.bt2"))
### Check if GTF annotation object exist or skip this step
### Check if the gtf object exists in the workspace
### Only run create_gtf_annotation if object doesnt exist
ws_gtf = annotation_gtf+"_GTF_Annotation"
ret = script_util.if_obj_exists(None,ws_client,params['ws_id'],"KBaseRNASeq.GFFAnnotation",[ws_gtf])
if not ret is None:
logger.info("GFF Annotation Exist for Genome Annotation {0}.... Skipping step ".format(annotation_gtf))
annot_name,annot_id = ret[0]
gtf_obj=ws_client.get_objects([{'ref' : annot_id}])[0]
gtf_id=gtf_obj['data']['handle']['id']
gtf_name=gtf_obj['data']['handle']['file_name']
try:
script_util.download_file_from_shock(logger, shock_service_url=self.urls['shock_service_url'], shock_id=gtf_id,filename=gtf_name, directory=tophat_dir,token=token)
token)
try:
logger.info("Unzipping Bowtie2 Indices")
script_util.unzip_files(logger,
os.path.join(bowtie2_dir, bw_index_files),
bowtie2_dir)
mv_dir = handler_util.get_dir(bowtie2_dir)
if mv_dir is not None:
script_util.move_files(logger, mv_dir, bowtie2_dir)
except Exception, e:
logger.error("".join(traceback.format_exc()))
raise Exception("Unzip indexfile error")
### Build Index for the fasta file
fasta_file = os.path.join(
bowtie2_dir,
handler_util.get_file_with_suffix(bowtie2_dir, ".fa") + ".fa")
bowtie2base = os.path.join(
bowtie2_dir, handler_util.get_file_with_suffix(bowtie2_dir, ".fa"))
bowtie2base_cmd = '{0} {1}'.format(fasta_file, bowtie2base)
try:
logger.info(
"Building Index for Hisat2 {0}".format(bowtie2base_cmd))
cmdline_output = script_util.runProgram(logger, "bowtie2-build",
bowtie2base_cmd, None,
bowtie2_dir)
except Exception, e:
raise Exception(
"Failed to run command {0}".format(bowtie2base_cmd))
### Check if GTF object exists in the workspace pull the gtf
ref_id = bowtie_index['data']['genome_id']
genome_name = ws_client.get_object_info_new(
def runEach(self, task_params):
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']
read_sample = task_params['job_id']
condition = task_params['label']
directory = task_params['tophat_dir']
ws_id = task_params['ws_id']
genome_id = task_params['annotation_id']
sampleset_id = task_params['sampleset_id']
gtf_file = task_params['gtf_file']
print "Downloading Read Sample{0}".format(read_sample)
logger.info("Downloading Read Sample{0}".format(read_sample))
try:
#r_sample = ws_client.get_objects(
# [{ 'name' : read_sample, 'workspace' : ws_id}])[0]
r_sample = script_util.ws_get_obj(logger, ws_client, ws_id,
read_sample)[0]
#r_sample_info = ws_client.get_object_info_new({"objects": [{'name': read_sample, 'workspace': ws_id}]})[0]
#sample_type = r_sample_info[2].split('-')[0]
sample_type = script_util.ws_get_type_name(logger, ws_client,
ws_id, read_sample)
sample_name = script_util.ws_get_obj_name4file(
self.logger, ws_client, ws_id, read_sample)
output_name = sample_name.split('.')[0] + "_tophat_alignment"
output_dir = os.path.join(directory, output_name)
#if not os.path.exists(output_dir): os.makedirs(output_dir)
#out_file = output_dir +"/accepted_hits.sam"
bowtie2_base = os.path.join(
directory,
handler_util.get_file_with_suffix(directory, ".rev.1.bt2"))
### Adding advanced options to Bowtie2Call
tophat_cmd = (' -p ' + str(self.num_threads))
if ('max_intron_length' in params
and params['max_intron_length'] is not None):
tophat_cmd += (' -I ' + str(params['max_intron_length']))
if ('min_intron_length' in params
and params['min_intron_length'] is not None):
tophat_cmd += (' -i ' + str(params['min_intron_length']))
if ('min_anchor_length' in params
and params['min_anchor_length'] is not None):
tophat_cmd += (' -a ' + str(params['min_anchor_length']))
if ('read_edit_dist' in params
and params['read_edit_dist'] is not None):
tophat_cmd += (' --read-edit-dist ' +
str(params['read_edit_dist']))
if ('read_gap_length' in params
and params['read_gap_length'] is not None):
tophat_cmd += (' --read-gap-length ' +
str(params['read_gap_length']))
if ('read_mismatches' in params
and params['read_mismatches'] is not None):
tophat_cmd += (' -N ' + str(params['read_mismatches']))
if ('library_type' in params
and params['library_type'] is not None):
tophat_cmd += (' --library-type ' + params['library_type'])
if ('report_secondary_alignments' in params
and int(params['report_secondary_alignments']) == 1):
tophat_cmd += ' --report-secondary-alignments'
if ('no_coverage_search' in params
and int(params['no_coverage_search']) == 1):
tophat_cmd += ' --no-coverage-search'
if ('preset_options' in params
and params['preset_options'] is not None):
tophat_cmd += ' --' + params['preset_options']
#out_file = output_dir +"/accepted_hits.sam"
try:
sample_ref = script_util.ws_get_ref(self.logger, ws_client,
ws_id, read_sample)
ds = script_util.ru_reads_download(self.logger, sample_ref,
directory, token)
except Exception, e:
self.logger.exception(e)
raise Exception(
"Unable to download reads file , {0}".format(read_sample))
if sample_type == 'KBaseAssembly.SingleEndLibrary' or sample_type == 'KBaseFile.SingleEndLibrary':
lib_type = 'SingleEnd'
tophat_cmd += ' -o {0} -G {1} {2} {3}'.format(
output_dir, gtf_file, bowtie2_base, ds['fwd'])
if sample_type == 'KBaseAssembly.PairedEndLibrary' or sample_type == 'KBaseFile.PairedEndLibrary':
lib_type = 'PairedEnd'
if sample_type == 'KBaseAssembly.PairedEndLibrary':
if ('orientation' in params
and params['orientation'] is not None):
tophat_cmd += (' --' + params['orientation'])
else:
# TODO: the following can be read from PEL object
if ('orientation' in params
and params['orientation'] is not None):
tophat_cmd += (' --' + params['orientation'])
tophat_cmd += ' -o {0} -G {1} {2} {3} {4}'.format(
output_dir, gtf_file, bowtie2_base, ds['fwd'], ds['rev'])
# if sample_type == 'KBaseAssembly.SingleEndLibrary' or sample_type == 'KBaseFile.SingleEndLibrary':
# lib_type = 'SingleEnd'
# if sample_type == 'KBaseAssembly.SingleEndLibrary':
# read_id = r_sample['data']['handle']['id']
# read_name = r_sample['data']['handle']['file_name']
# else:
# read_id = r_sample['data']['lib']['file']['id']
# read_name = r_sample['data']['lib']['file']['file_name']
# try:
# script_util.download_file_from_shock(self.logger, shock_service_url=self.urls['shock_service_url'], shock_id=read_id,filename=read_name, directory=directory,token=token)
# tophat_cmd += ' -o {0} -G {1} {2} {3}'.format(output_dir,gtf_file,bowtie2_base,os.path.join(directory,read_name))
# except Exception,e:
# self.logger.exception(e)
# raise Exception( "Unable to download shock file , {0}".format(read_name))
# if sample_type == 'KBaseAssembly.PairedEndLibrary' or sample_type == 'KBaseFile.PairedEndLibrary':
# lib_type = 'PairedEnd'
# if sample_type == 'KBaseAssembly.PairedEndLibrary':
# if('orientation' in params and params['orientation'] is not None): tophat_cmd += ( ' --'+params['orientation'])
# read1_id = r_sample['data']['handle_1']['id']
# read1_name = r_sample['data']['handle_1']['file_name']
# read2_id = r_sample['data']['handle_2']['id']
# read2_name = r_sample['data']['handle_2']['file_name']
# else:
# # TODO: the following can be read from PEL object
# if('orientation' in params and params['orientation'] is not None): tophat_cmd += ( ' --'+params['orientation'])
# read1_id = r_sample['data']['lib1']['file']['id']
# read1_name = r_sample['data']['lib1']['file']['file_name']
# read2_id = r_sample['data']['lib2']['file']['id']
# read2_name = r_sample['data']['lib2']['file']['file_name']
# try:
# script_util.download_file_from_shock(self.logger, shock_service_url=self.urls['shock_service_url'], shock_id=read1_id,filename=read1_name, directory=directory,token=token)
# script_util.download_file_from_shock(self.logger, shock_service_url=self.urls['shock_service_url'], shock_id=read2_id,filename=read2_name, directory=directory,token=token)
# tophat_cmd += ' -o {0} -G {1} {2} {3} {4}'.format(output_dir,gtf_file,bowtie2_base,os.path.join(directory,read1_name),os.path.join(directory,read2_name))
# except Exception,e:
# raise Exception( "Unable to download shock file , {0} or {1}".format(read1_name,read2_name))
try:
self.logger.info("Executing: tophat {0}".format(tophat_cmd))
cmdline_output, cmd_err = script_util.runProgram(
self.logger, "tophat", tophat_cmd, None, directory)
except Exception, e:
raise Exception("Failed to run command {0}\n{1}\n{2}".format(
tophat_cmd, cmdline_output, cmd_err))