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_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)
def run(self):
'''Runs recovery from start to finish using command line arguments.'''
args = self.get_args()
self.test = args.test
self.ignore = False
if args.ignore_properties:
print "Ignoring DXFile properties (will post to test server)"
self.ignore = args.ignore_properties
self.server_key = 'test' # mandated because option is dangerous
self.server_key = args.server
self.authid, self.authpw, self.server = dxencode.processkey(self.server_key)
if self.server_key == "www":
self.acc_prefix = "ENCFF"
self.proj_name = dxencode.env_get_current_project()
if self.proj_name == None or args.project != None:
self.proj_name = args.project
if self.proj_name == None:
print "Please enter a '--project' to run in."
sys.exit(1)
self.project = dxencode.get_project(self.proj_name)
self.proj_id = self.project.get_id()
print "== Running in project [%s] and will attempt recovery to the [%s] server ==" % \
(self.proj_name,self.server_key)
exp_count = 0
halted = 0
total_recovered = 0
for exp_id in args.experiments:
sys.stdout.flush() # Slow running job should flush to piped log
self.exp_id = exp_id
self.obj_cache["exp"] = {} # clear exp cache, which will hold exp specific wf_run and step_run objects
# 1) Lookup experiment type from encoded, based on accession
print "Working on %s..." % self.exp_id
self.exp = dxencode.get_exp(self.exp_id,must_find=True,key=self.server_key)
if self.exp == None or self.exp["status"] == "error":
print "Unable to locate experiment %s in encoded (%s)" % (self.exp_id, self.server_key)
continue
self.exp_type = dxencode.get_exp_type(self.exp_id,self.exp,self.EXPERIMENT_TYPES_SUPPORTED)
if self.exp_type == None:
continue
# 2) Locate the experiment accession named folder
# NOTE: genome and annotation are not known for this exp yet, so the umbrella folder is just based on exp_type
self.umbrella_folder = dxencode.umbrella_folder(args.folder,self.FOLDER_DEFAULT,self.proj_name,self.exp_type)
self.exp_folder = dxencode.find_exp_folder(self.project,exp_id,self.umbrella_folder,warn=True)
if self.exp_folder == None:
continue
print "- Examining %s:%s for '%s' results..." % \
(self.proj_name, self.exp_folder, self.exp_type)
# 3) Given the experiment type, determine the expected results
self.pipeline = self.pipeline_specification(args,self.exp_type,self.exp_folder)
self.replicates = self.find_replicate_folders(self.exp_folder, verbose=args.verbose)
# 4) Given expected results locate any files (by glob) that should be posted for
# a) each single replicate (in replicate sub-folders named as reN_N/
# b) combined replicates in the experiment folder itself
files_expected = self.find_expected_files(self.exp_folder, self.replicates, verbose=args.verbose)
print "- Found %d files that are available in DX." % len(files_expected)
if len(files_expected) == 0:
continue
# 5) For each file that should be posted, determine if the file needs to be posted.
files_posted = self.find_posted_files(files_expected, test=self.test, verbose=args.verbose) #True)
print "- Found %d files that have been posted" % len(files_posted)
if len(files_posted) == 0:
continue
# 6) For each file that needs to be posted:
exp_count += 1
file_count = 0
recovery_count = 0
for (out_type,rep_tech,fid) in files_posted:
sys.stdout.flush() # Slow running job should flush to piped log
accession = self.found[fid]['accession']
file_name = dxencode.file_path_from_fid(fid)
if args.start_at != None:
if accession != args.start_at and not file_name.endswith(args.start_at):
continue
else:
print "- Starting at %s" % (file_name)
args.start_at = None
# a) discover all necessary dx information needed for post.
# b) gather any other information necessary from dx and encoded.
print "- Handle file %s %s" % (accession,dxencode.file_path_from_fid(fid))
payload = self.make_payload_obj(out_type,rep_tech,fid, verbose=args.verbose)
file_count += 1
# c) Update encoded database only if necessary.
if self.file_metadata_recovery(fid,payload,args.test,verbose=args.verbose):
recovery_count += 1
if args.files != 0 and file_count >= args.files: # Short circuit for test
print "- Just trying %d file(s) by request" % file_count
break
if not args.test:
print "- For %s Processed %d file(s), recovered %s" % (self.exp_id, file_count, recovery_count)
else:
print "- For %s Processed %d file(s), would recover %s" % (self.exp_id, file_count, recovery_count)
total_recovered += recovery_count
if not args.test:
print "Processed %d experiment(s), halted %d, recovered %d file(s)" % (exp_count, halted, total_recovered)
else:
print "Processed %d experiment(s), halted %d, would recover %d file(s)" % (exp_count, halted, total_recovered)
if halted == exp_count:
sys.exit(1)
print "(finished)"
def run(self):
'''Override super.run()'''
args = self.get_args()
self.test = args.test
self.server_key = args.server
if self.server_key != "test":
self.acc_prefix = "ENCFF"
self.proj_name = dxencode.env_get_current_project()
if self.proj_name == None or args.project != None:
self.proj_name = args.project
if self.proj_name == None:
print "Please enter a '--project' to run in."
sys.exit(1)
self.project = dxencode.get_project(self.proj_name)
self.proj_id = self.project.get_id()
print "== Running in project [%s] and will post to the [%s] server ==" % \
(self.proj_name,self.server_key)
exp_count = 0
halted = 0
total_posted = 0
for exp_id in args.experiments:
sys.stdout.flush() # Slow running job should flush to piped log
# 1) Lookup experiment type from encoded, based on accession
print "Working on %s..." % exp_id
self.exp = dxencode.get_exp(exp_id,
must_find=False,
key=self.server_key)
if self.exp == None or self.exp["status"] == "error":
print "Unable to locate experiment %s in encoded" % exp_id
continue
self.exp_type = self.get_exp_type(exp_id)
if self.exp_type == None:
continue
# 2) Locate the experiment accession named folder
self.exp_folder = dxencode.find_exp_folder(self.project,
exp_id,
args.results_folder,
warn=True)
if self.exp_folder == None:
continue
print "- Examining %s:%s for '%s' results..." % \
(self.proj_name, self.exp_folder, self.exp_type)
# 3) Given the experiment type, determine the expected results
self.pipeline = self.pipeline_specification(
args, self.exp_type, self.exp_folder)
self.replicates = self.find_replicate_folders(self.exp_folder,
verbose=args.verbose)
# 4) Given expected results locate any files (by glob) that should be posted for
# a) each single replicate (in replicate sub-folders named as reN_N/
# b) combined replicates in the experiment folder itself
files_expected = self.find_expected_files(self.exp_folder,
self.replicates,
verbose=args.verbose)
print "- Found %d files that are available to post." % len(
files_expected)
if len(files_expected) == 0:
continue
# 5) For each file that should be posted, determine if the file needs to be posted.
files_to_post = {
x[2]: x
for x in self.find_needed_files(files_expected,
verbose=args.verbose)
}
# index on dx file id
print "- Found %d files that need to be posted" % len(
files_to_post.keys())
# 6) For each file that needs to be posted:
exp_count += 1
file_count = 0
post_count = 0
for (out_type, rep_tech, fid) in files_expected:
sys.stdout.flush(
) # Slow running job should flush to piped log
# a) discover all necessary dx information needed for post.
# b) gather any other information necessary from dx and encoded.
print " Handle file %s" % dxencode.file_path_from_fid(fid)
job = dxencode.job_from_fid(fid)
try:
derived_from = self.find_derived_from(
fid, job, args.verbose)
except dxpy.exceptions.ResourceNotFound, e:
print "WARN: derived_from failed %s" % e
derived_from = []
if not files_to_post.get(fid, ()):
f_obj = self.found.get(fid, None)
if f_obj:
current_derived_from = f_obj['derived_from']
if derived_from and not current_derived_from:
print "Need to patch derived_from for %s/%s to %s (currently: %s)" % (
f_obj['accession'], fid, derived_from,
current_derived_from)
else:
print "Derived from for %s good" % f_obj[
'accession']
else:
print "File %s (%s) from %s/%s not found @ DNANexus" % (
fid, out_type, exp_id, rep_tech)
#POSTING
else:
payload = self.make_payload_obj(out_type,
rep_tech,
fid,
verbose=args.verbose)
if args.force_annotation:
print "WARN: forcing genome_annotation to be %s" % args.force_annotation
payload['genome_annotation'] = args.force_annotation
file_count += 1
# c) Post file and update encoded database.
accession = self.file_post(fid, payload, args.test)
if accession == None:
print "* HALTING %s - post failure could compromise 'derived_from'" % \
(self.exp_id)
halted += 1
break
# d) Update dnanexus file with file accession tag.
if not args.test:
post_count += 1
self.file_mark_accession(fid, accession, args.test)
print "- For %s Processed %d file(s), posted %s" % \
(self.exp_id, file_count, post_count)
total_posted += post_count
def file_metadata_recovery(self,fid,payload,test=True,verbose=False):
'''Compares DX and encoded metadata and updates ENCODEd is necessary.'''
recovered = False
if fid not in self.found:
print "* ERROR: Expecting to have exp_file for " + dxencode.file_path_from_fid(fid,projectToo=True)
sys.exit(1)
enc_file = self.found[fid]
accession = enc_file['accession']
update_payload = {}
patch_required = False
# Compare derived from and update if necessary
append_derived_by = False
if self.APPEND_FLAG in payload['derived_from']:
payload['derived_from'].remove(self.APPEND_FLAG)
append_derived_by = True
derived_diffs = len(payload['derived_from'])
if verbose:
print >> sys.stderr, "> DX file derived_from:"
print >> sys.stderr, json.dumps(payload['derived_from'],indent=4)
enc_derived = enc_file.get('derived_from')
if enc_derived != None:
if verbose:
print >> sys.stderr, "> Enc file derived_from:"
print >> sys.stderr, json.dumps(enc_derived,indent=4)
if append_derived_by or len(enc_derived) == derived_diffs:
for acc in payload['derived_from']:
for inp_file in enc_derived:
if acc == inp_file.get('accession'):
derived_diffs -= 1
else:
if verbose:
print >> sys.stderr, "Enc file derived_from: Not Found"
if derived_diffs > 0:
if append_derived_by:
print " + Need to append to 'derived_from'."
if enc_derived != None:
for inp_file in enc_derived:
acc = inp_file.get('accession')
if acc != None and acc not in payload['derived_from']:
payload['derived_from'].append(acc)
else:
print " + Need to update 'derived_from'."
update_payload['derived_from'] = payload['derived_from']
patch_required = True
# Compare wf_run/step_run and update if necessary
step_run_diff = True
dx_step_run = payload.get("step_run")
if dx_step_run != None:
dx_step_run = dx_step_run.split('/')[-1] # Just the alias please
if verbose:
print >> sys.stderr, "> DX file step_run:" + dx_step_run
else:
print "* ERROR: payload is missing 'step_run'"
enc_step_run = enc_file.get("step_run")
if verbose:
if enc_step_run != None:
print >> sys.stderr, "> Enc file step_run:" + enc_step_run
#print >> sys.stderr, json.dumps(enc_step_run,indent=4)
else:
print >> sys.stderr, "> Enc file step_run: Not Found"
if dx_step_run == None and enc_step_run == None:
step_run_diff = False
elif dx_step_run != None and enc_step_run != None:
step_run = self.enc_lookup_json(enc_step_run,must_find=True)
if verbose:
print >> sys.stderr, "> Actual step_run:"
print >> sys.stderr, json.dumps(step_run,indent=4)
aliases = step_run.get('aliases')
if aliases != None:
if verbose:
print >> sys.stderr, "> aliass:"
print >> sys.stderr, json.dumps(aliases,indent=4)
if dx_step_run in aliases:
step_run_diff = False
if step_run_diff:
print " + Need to update 'step_run'."
update_payload["step_run"] = payload["step_run"]
patch_required = True
# What about lab?
if 'lab' not in enc_file:
print " + Need to update missing 'lab' new: '"+payload['lab']+"'"
update_payload['lab'] = payload['lab']
patch_required = True
else:
lab = enc_file.get('lab')
if '@id' not in lab or lab['@id'] != payload['lab']:
print " + Need to update 'lab' new: '"+payload['lab']+"'"
#print json.dumps(lab,indent=4,sort_keys=True)
update_payload['lab'] = payload['lab']
patch_required = True
# award?
if 'award' not in enc_file:
print " + Need to update missing 'award' new: '"+payload['award']+"'"
update_payload['award'] = payload['award']
patch_required = True
elif enc_file['award'] != payload['award']:
print " + Need to update 'award' new: '"+payload['award']+"' enc: '"+enc_file['award']+"'"
update_payload['award'] = payload['award']
patch_required = True
if patch_required:
update_payload['notes'] = payload['notes'] # Only update notes if other things update, then always update notes
if test:
print " * Would patch file: '%s'" % dxencode.file_path_from_fid(fid)
#print json.dumps(update_payload,indent=4,sort_keys=True)
recovered = True
else:
try:
ret = dxencode.encoded_patch_obj(accession, update_payload, self.server, self.authid, self.authpw)
except:
print "Failed to patch file: '%s'" % dxencode.file_path_from_fid(fid)
print json.dumps(update_payload,indent=4,sort_keys=True)
sys.exit(1)
print " * Patched file: '%s'" % dxencode.file_path_from_fid(fid)
recovered = True
else:
print " - No need to patch file."
return recovered
