def find_ref_files(self,priors):
'''Locates all reference files based upon gender, organism and annotation.'''
topIx = self.psv['refLoc']+self.REFERENCE_FILES['tophat_index'][self.psv['genome']][self.psv['gender']][self.psv['annotation']]
topIxFid = dxencode.find_file(topIx,dxencode.REF_PROJECT_DEFAULT)
if topIxFid == None:
sys.exit("ERROR: Unable to locate TopHat index file '" + topIx + "'")
else:
priors['tophat_index'] = topIxFid
starIx = self.psv['refLoc']+self.REFERENCE_FILES['star_index'][self.psv['genome']][self.psv['gender']][self.psv['annotation']]
starIxFid = dxencode.find_file(starIx,dxencode.REF_PROJECT_DEFAULT)
if starIxFid == None:
sys.exit("ERROR: Unable to locate STAR index file '" + starIx + "'")
else:
priors['star_index'] = starIxFid
rsemIx = self.psv['refLoc']+self.REFERENCE_FILES['rsem_index'][self.psv['genome']][self.psv['annotation']]
rsemIxFid = dxencode.find_file(rsemIx,dxencode.REF_PROJECT_DEFAULT)
if rsemIxFid == None:
sys.exit("ERROR: Unable to locate RSEM index file '" + rsemIx + "'")
else:
priors['rsem_index'] = rsemIxFid
chromSizes = self.psv['refLoc']+self.REFERENCE_FILES['chrom_sizes'][self.psv['genome']][self.psv['gender']]
chromSizesFid = dxencode.find_file(chromSizes,dxencode.REF_PROJECT_DEFAULT)
if chromSizesFid == None:
sys.exit("ERROR: Unable to locate Chrom Sizes file '" + chromSizes + "'")
else:
priors['chrom_sizes'] = chromSizesFid
self.psv['ref_files'] = self.REFERENCE_FILES.keys()
def find_ref_files(self,priors):
'''Locates all reference files based upon organism and gender.'''
star_ix = self.psv['refLoc']+self.REFERENCE_FILES['star_index'][self.psv['genome']][self.psv['gender']]
star_ix_fid = dxencode.find_file(star_ix,dxencode.REF_PROJECT_DEFAULT)
if star_ix_fid == None:
sys.exit("ERROR: Unable to locate STAR index file '" + star_ix + "'")
else:
priors['star_index'] = star_ix_fid
chrom_sizes = self.psv['refLoc']+self.REFERENCE_FILES['chrom_sizes'][self.psv['genome']][self.psv['gender']]
chrom_sizes_fid = dxencode.find_file(chrom_sizes,dxencode.REF_PROJECT_DEFAULT)
if chrom_sizes_fid == None:
sys.exit("ERROR: Unable to locate Chrom Sizes file '" + chrom_sizes + "'")
else:
priors['chrom_sizes'] = chrom_sizes_fid
self.psv['ref_files'] = self.REFERENCE_FILES.keys()
def find_control_file(self,rep,default=None):
'''Attempts to find an appropriate control file.'''
# TODO Make more generic and move to dxencode.py when needed.
(AUTHID,AUTHPW,SERVER) = dxencode.processkey(self.server_key)
if 'controls' not in rep:
return None
for file_key in rep['controls']:
if isinstance(file_key,list):
file_key = file_key[0]
file_obj = dxencode.enc_lookup_json(file_key,self.server_key,frame='embedded')
rep_id = file_obj["replicate"]['@id']
rep_obj = dxencode.enc_lookup_json(rep_id,self.server_key,frame='embedded')
exp_id = rep_obj['experiment']['@id'].split('/')[2]
rep_tech = "rep%s_%s" % \
(rep_obj['biological_replicate_number'], rep_obj['technical_replicate_number'])
control_root = self.psv['control_path']
# Cheating:
if self.proj_name == "scratchPad" and self.psv['control_path'] == self.CONTROL_ROOT_FOLDER:
control_root = "/lrna"
path_n_glob = control_root + exp_id + '/' + rep_tech + '/' + self.CONTROL_FILE_GLOB
target_folder = dxencode.find_folder(exp_id + '/' + rep_tech,self.project,control_root)
#print "Target found [%s]" % target_folder
if target_folder != None:
path_n_glob = target_folder + '/' + self.CONTROL_FILE_GLOB
fid = dxencode.find_file(path_n_glob,self.proj_id,multiple=False,recurse=False)
if fid != None:
return dxencode.file_path_from_fid(fid)
if default != None:
return default
#print json.dumps(rep,indent=4)
print "Unable to find control in search of %s" % rep['controls']
sys.exit(1)
def find_ref_files(self,priors):
'''Locates all reference files based upon organism and gender.'''
#bwaIx = self.psv['refLoc']+self.REFERENCE_FILES['bwa_index'][self.psv['genome']][self.psv['gender']]
base_dir = '/' + self.psv['genome'] + "/dna-me/"
dmeIx = base_dir+self.REFERENCE_FILES["dme_ix"][self.psv['genome']]
dmeIxFid = dxencode.find_file(dmeIx,dxencode.REF_PROJECT_DEFAULT)
if dmeIxFid == None:
sys.exit("ERROR: Unable to locate Bismark index file '" + dmeIx + "'")
else:
priors['dme_ix'] = dmeIxFid
self.psv['ref_files'] = self.REFERENCE_FILES.keys()
def find_ref_files(priors, psv):
'''Locates all reference files based upon gender, organism and annotation.'''
refFiles = {}
topIx = psv['refLoc'] + GENOME_REFERENCES['tophat_index'][psv['genome']][
psv['gender']][psv['annotation']]
topIxFid = dxencode.find_file(topIx, dxencode.REF_PROJECT_DEFAULT)
if topIxFid == None:
sys.exit("ERROR: Unable to locate TopHat index file '" + topIx + "'")
else:
priors['tophat_index'] = topIxFid
starIx = psv['refLoc'] + GENOME_REFERENCES['star_index'][psv['genome']][
psv['gender']][psv['annotation']]
starIxFid = dxencode.find_file(starIx, dxencode.REF_PROJECT_DEFAULT)
if starIxFid == None:
sys.exit("ERROR: Unable to locate STAR index file '" + starIx + "'")
else:
priors['star_index'] = starIxFid
rsemIx = psv['refLoc'] + GENOME_REFERENCES['rsem_index'][psv['genome']][
psv['annotation']]
rsemIxFid = dxencode.find_file(rsemIx, dxencode.REF_PROJECT_DEFAULT)
if rsemIxFid == None:
sys.exit("ERROR: Unable to locate RSEM index file '" + rsemIx + "'")
else:
priors['rsem_index'] = rsemIxFid
chromSizes = psv['refLoc'] + GENOME_REFERENCES['chrom_sizes'][
psv['genome']][psv['gender']]
chromSizesFid = dxencode.find_file(chromSizes,
dxencode.REF_PROJECT_DEFAULT)
if chromSizesFid == None:
sys.exit("ERROR: Unable to locate Chrom Sizes file '" + chromSizes +
"'")
else:
priors['chrom_sizes'] = chromSizesFid
psv['ref_files'] = GENOME_REFERENCES.keys()
def find_control_file(self,rep,default=None):
'''Attempts to find an appropriate control file.'''
# TODO Make more generic and move to dxencode.py when needed.
(AUTHID,AUTHPW,SERVER) = dxencode.processkey(self.server_key)
for file_key in rep['controls']:
url = '%s%s/?format=json&frame=embedded' % (SERVER,file_key)
#print '-- ' + AUTHID + " " + AUTHPW + " " + SERVER + " " + url
try:
response = dxencode.encoded_get(url, AUTHID, AUTHPW)
file_obj = response.json()
except:
print "URL to control [%s] returned ?" % url
print response
sys.exit(1)
#print json.dumps(response,indent=4)
rep_id = file_obj["replicate"]['@id']
url = '%s%s/?format=json&frame=embedded' % (SERVER,rep_id)
try:
response = dxencode.encoded_get(url, AUTHID, AUTHPW)
rep_obj = response.json()
except:
print "URL to replicate [%s] returned ?" % url
print response
sys.exit(1)
exp_id = rep_obj['experiment'].split('/')[2]
rep_tech = "rep%s_%s" % \
(rep_obj['biological_replicate_number'], rep_obj['technical_replicate_number'])
# default by cheating
if self.proj_name == dxencode.PRODUCTION_PROJECT:
control_root = "/long-RNA-seq/runs/"
else:
control_root = self.CONTROL_ROOT_FOLDER
path_n_glob = control_root + exp_id + '/' + rep_tech + '/' + self.CONTROL_FILE_GLOB
target_folder = dxencode.find_folder(exp_id + '/' + rep_tech,self.project,control_root)
#print "Target found [%s]" % target_folder
if target_folder != None:
path_n_glob = target_folder + '/' + self.CONTROL_FILE_GLOB
fid = dxencode.find_file(path_n_glob,self.proj_id,multiple=False,recurse=False)
if fid != None:
return dxencode.file_path_from_fid(fid)
if default != None:
return default
print "Unable to find control in search of %s" % rep['controls']
sys.exit(1)
