def _manifest_upload(syn, df):
items = []
for i, row in df.iterrows():
file = File(
path=row['path'],
parent=row['parent'],
**{key: row[key]
for key in FILE_CONSTRUCTOR_FIELDS if key in row},
)
file.annotations = dict(
row.drop(FILE_CONSTRUCTOR_FIELDS + STORE_FUNCTION_FIELDS +
REQUIRED_FIELDS + PROVENANCE_FIELDS,
errors='ignore'))
item = _SyncUploadItem(
file,
row['used'] if 'used' in row else [],
row['executed'] if 'executed' in row else [],
{key: row[key]
for key in STORE_FUNCTION_FIELDS if key in row},
)
items.append(item)
with _sync_executor(syn) as executor:
uploader = _SyncUploader(syn, executor)
uploader.upload(items)
return True
def upload(args,syn):
if args.dataType == "rnaseq":
parentId = "syn6034916"
pipeline = "syn6126122"
dataType = "RNASeq"
elif args.dataType == "dnaseq":
parentId = "syn6034751"
pipeline = "syn6126123"
dataType = "TargDNASeq"
elif args.dataType == "snparray":
parentId = "syn6038475"
pipeline = "syn6126121"
dataType = "SNParray"
elif args.dataType == "exparray":
parentId = "syn6038915"
pipeline = "syn6126120"
dataType = "expression_microarray"
elif args.dataType == "exome":
parentId = "syn6115597"
dataType = "exome"
pipeline = ""
else:
raise ValueError("dataType needs to be rnaseq/dnaseq/snparray/exparray/exome")
if args.workflow is not None:
workflow = syn.get(pipeline,downloadFile=False)
workflow.path = args.workflow
workflow.name = os.path.basename(args.workflow)
workflow = syn.store(workflow)
pipeline = workflow.id
fileEnt = File(args.input,parent=parentId)
fileEnt.annotations = temp.to_dict('index').values()[0]
fileEnt.dataType = dataType
fileEnt.sampleId = sampleId
fileEnt = syn.store(fileEnt,used = pipeline)
return(fileEnt.id)
def _manifest_upload(syn, df):
for i, row in df.iterrows():
#Todo extract known constructor variables
kwargs = {key: row[key] for key in FILE_CONSTRUCTOR_FIELDS if key in row }
entity = File(row['path'], parent=row['parent'], **kwargs)
entity.annotations = dict(row.drop(FILE_CONSTRUCTOR_FIELDS+STORE_FUNCTION_FIELDS+REQUIRED_FIELDS, errors = 'ignore'))
#Update provenance list again to replace all file references that were uploaded
if 'used' in row:
row['used'] = syn._convertProvenanceList(row['used'])
if 'executed' in row:
row['executed'] = syn._convertProvenanceList(row['executed'])
kwargs = {key: row[key] for key in STORE_FUNCTION_FIELDS if key in row}
entity = syn.store(entity, **kwargs)
return True
def create_entity(syn: Synapse, name: str, link: str,
annotations: dict) -> File:
"""Creates evaluation queue
Args:
name: Name of queue
Returns:
a synapseclient.Evaluation
"""
file_ent = File(name=name,
path=link,
parentId="syn21897226",
synapseStore=False)
file_ent.annotations = annotations
return syn.store(file_ent)
def upload_fastq(fastq, bid, annot, theFilepath):
print 'Uploading {}'.format(fastq)
barcode = '_'.join(fastq.split('_')[4:7])
platform = annot['platform']
filename = synapse_filename_template.format(bid=bid,
platform=platform,
barcode=barcode)
fastq_synapse = File(theFilepath,
name=filename,
parent=FASTQ_FOLDER_SYNAPSE)
# Deprecated. Will error out trying to upload_fastq
# fastq_synapse.properties.fileNameOverride = filename
fastq_synapse.annotations = annot
print "Uploading"
syn.store(fastq_synapse)
def upload_file(self, file_path, overwrite=False):
"""Saves file to Synapse"""
fp = file_path['file_p']
ap = _get_anno_p(fp)
local_md5 = calc_content_md5(fp)
parent = self.get_or_create_folder(fp.parent, self.parent_synid)
file = File(path=str(fp), parent=parent)
annotations = load_json(ap)
annotations = {
k.replace('[', 'LFTB').replace(']', 'RGTB'): v
for k, v in annotations.items()
}
file.annotations = annotations
file_r = self.get_or_create_entity(file, returnid=False)
if overwrite:
print(local_md5)
print(file_r.md5)
if local_md5 != file_r.md5:
print('mismatch. overwriting.')
self.syn.store(file)
# Get parentFolder
def getFileFolder(fileName):
if fileName.endswith(('.fastq', '.trimmed.fq')):
return "raw"
if fileName.endswith(('.bam', '.bai', '.unmapped.fq')):
return "aligned"
if fileName.endswith(('.readsLength.freq', '.txt', '.16_25nt.fq',
'.miRanalyzer.zip')) or re.search(r'_vs_', fileName):
return "analyzed"
return "miRNA"
f['fileFolder'] = f['fileName'].apply(lambda x: getFileFolder(x))
# Get the folder synapse Ids
with open(args.folderInfo) as info:
folderIds = yaml.load(info)
# Link files
for index, row in f.iterrows():
entityFile = File(parentId=folderIds[row['fileFolder']],
name=row['fileName'])
entityFile.annotations = dict(assay=row['assay'],
fileFormat=row['fileFormat'],
cellLine=row['cellLine'])
newEntity = storePSON(entityFile,
row['filePath'],
contentSize=row['fileSize'],
md5=row['md5'],
syn=syn)
def _copyFile(syn,
entity,
destinationId,
version=None,
updateExisting=False,
setProvenance="traceback",
skipCopyAnnotations=False):
"""
Copies most recent version of a file to a specified synapse ID.
:param entity: A synapse ID of a File entity
:param destinationId: Synapse ID of a folder/project that the file wants to be copied to
:param version: Can specify version of a file.
Default to None
:param updateExisting: Can choose to update files that have the same name
Default to False
:param setProvenance: Has three values to set the provenance of the copied entity:
traceback: Sets to the source entity
existing: Sets to source entity's original provenance (if it exists)
None: No provenance is set
:param skipCopyAnnotations: Skips copying the annotations
Default is False
"""
ent = syn.get(entity,
downloadFile=False,
version=version,
followLink=False)
# CHECK: If File is in the same parent directory (throw an error) (Can choose to update files)
if not updateExisting:
existingEntity = syn.findEntityId(ent.name, parent=destinationId)
if existingEntity is not None:
raise ValueError(
'An entity named "%s" already exists in this location. File could not be copied'
% ent.name)
profile = syn.getUserProfile()
# get provenance earlier to prevent errors from being called in the end
# If traceback, set activity to old entity
if setProvenance == "traceback":
act = Activity("Copied file", used=ent)
# if existing, check if provenance exists
elif setProvenance == "existing":
try:
act = syn.getProvenance(ent.id)
except SynapseHTTPError as e:
if e.response.status_code == 404:
act = None
else:
raise e
elif setProvenance is None or setProvenance.lower() == 'none':
act = None
else:
raise ValueError(
'setProvenance must be one of None, existing, or traceback')
# Grab entity bundle
bundle = syn._getEntityBundle(ent.id,
version=ent.versionNumber,
requestedObjects={
'includeEntity': True,
'includeFileHandles': True
})
fileHandle = synapseclient.core.utils.find_data_file_handle(bundle)
createdBy = fileHandle['createdBy']
# CHECK: If the user created the file, copy the file by using fileHandleId else copy the fileHandle
if profile.ownerId == createdBy:
newdataFileHandleId = ent.dataFileHandleId
else:
copiedFileHandle = copyFileHandles(syn, [fileHandle], ["FileEntity"],
[bundle['entity']['id']],
[fileHandle['contentType']],
[fileHandle['fileName']])
# Check if failurecodes exist
copyResult = copiedFileHandle[0]
if copyResult.get("failureCode") is not None:
raise ValueError("%s dataFileHandleId: %s" %
(copyResult["failureCode"],
copyResult['originalFileHandleId']))
newdataFileHandleId = copyResult['newFileHandle']['id']
new_ent = File(dataFileHandleId=newdataFileHandleId,
name=ent.name,
parentId=destinationId)
# Set annotations here
if not skipCopyAnnotations:
new_ent.annotations = ent.annotations
# Store provenance if act is not None
if act is not None:
new_ent = syn.store(new_ent, activity=act)
else:
new_ent = syn.store(new_ent)
# Leave this return statement for test
return new_ent['id']
f['fileFolder'] = f['filePath'].apply(lambda x: getFileFolder(x))
f.fillna('None', inplace=True)
# Get the folder synapse Ids
with open(args.folderInfo) as info:
folderIds = yaml.load(info)
# Link files
for index, row in f.iterrows():
entityFile = File(parentId=folderIds[row['fileFolder']],
name=row['fileName'])
annotations = dict(assay=row['assay'],
fileFormat=row['fileFormat'],
cellLine=row['cellLine'],
organ=row['organ'],
diagnosis=row['diagnosis'],
cellType=row['cellType'],
tumorType=row['tumorType'],
catalogNumber=row['catalogNumber'],
dataType="geneExpression",
consortium="PSON")
entityFile.annotations = dict(
(k, v) for k, v in annotations.iteritems() if v != 'None')
newEntity = storePSON(entityFile,
row['filePath'],
contentSize=row['fileSize'],
md5=row['md5'],
syn=syn)
def _copyFile(syn, entity, destinationId, version=None, updateExisting=False, setProvenance="traceback",
skipCopyAnnotations=False):
"""
Copies most recent version of a file to a specified synapse ID.
:param entity: A synapse ID of a File entity
:param destinationId: Synapse ID of a folder/project that the file wants to be copied to
:param version: Can specify version of a file.
Default to None
:param updateExisting: Can choose to update files that have the same name
Default to False
:param setProvenance: Has three values to set the provenance of the copied entity:
traceback: Sets to the source entity
existing: Sets to source entity's original provenance (if it exists)
None: No provenance is set
:param skipCopyAnnotations: Skips copying the annotations
Default is False
"""
ent = syn.get(entity, downloadFile=False, version=version, followLink=False)
# CHECK: If File is in the same parent directory (throw an error) (Can choose to update files)
if not updateExisting:
existingEntity = syn.findEntityId(ent.name, parent=destinationId)
if existingEntity is not None:
raise ValueError('An entity named "%s" already exists in this location. File could not be copied'
% ent.name)
profile = syn.getUserProfile()
# get provenance earlier to prevent errors from being called in the end
# If traceback, set activity to old entity
if setProvenance == "traceback":
act = Activity("Copied file", used=ent)
# if existing, check if provenance exists
elif setProvenance == "existing":
try:
act = syn.getProvenance(ent.id)
except SynapseHTTPError as e:
if e.response.status_code == 404:
act = None
else:
raise e
elif setProvenance is None or setProvenance.lower() == 'none':
act = None
else:
raise ValueError('setProvenance must be one of None, existing, or traceback')
# Grab entity bundle
bundle = syn._getEntityBundle(ent.id, version=ent.versionNumber, bitFlags=0x800 | 0x1)
fileHandle = synapseclient.utils.find_data_file_handle(bundle)
createdBy = fileHandle['createdBy']
# CHECK: If the user created the file, copy the file by using fileHandleId else copy the fileHandle
if profile.ownerId == createdBy:
newdataFileHandleId = ent.dataFileHandleId
else:
copiedFileHandle = copyFileHandles(syn, [fileHandle], ["FileEntity"], [bundle['entity']['id']],
[fileHandle['contentType']], [fileHandle['fileName']])
# Check if failurecodes exist
copyResult = copiedFileHandle['copyResults'][0]
if copyResult.get("failureCode") is not None:
raise ValueError("%s dataFileHandleId: %s" % (copyResult["failureCode"],
copyResult['originalFileHandleId']))
newdataFileHandleId = copyResult['newFileHandle']['id']
new_ent = File(dataFileHandleId=newdataFileHandleId, name=ent.name, parentId=destinationId)
# Set annotations here
if not skipCopyAnnotations:
new_ent.annotations = ent.annotations
# Store provenance if act is not None
if act is not None:
new_ent = syn.store(new_ent, activity=act)
else:
new_ent = syn.store(new_ent)
# Leave this return statement for test
return new_ent['id']
def _copyFile(syn, entity, destinationId, version=None, update=False, setProvenance="traceback"):
"""
Copies most recent version of a file to a specified synapse ID.
:param entity: A synapse ID of a File entity
:param destinationId: Synapse ID of a folder/project that the file wants to be copied to
:param version: Can specify version of a file.
Default to None
:param update: Can choose to update files that have the same name
Default to False
:param setProvenance: Has three values to set the provenance of the copied entity:
traceback: Sets to the source entity
existing: Sets to source entity's original provenance (if it exists)
None: No provenance is set
"""
ent = syn.get(entity, downloadFile=False, version=version, followLink=False)
#CHECK: If File is in the same parent directory (throw an error) (Can choose to update files)
if not update:
search = syn.query('select name from entity where parentId =="%s"'%destinationId)
for i in search['results']:
if i['entity.name'] == ent.name:
raise ValueError('An item named "%s" already exists in this location. File could not be copied'%ent.name)
profile = syn.getUserProfile()
# get provenance earlier to prevent errors from being called in the end
# If traceback, set activity to old entity
if setProvenance == "traceback":
act = Activity("Copied file", used=ent)
# if existing, check if provenance exists
elif setProvenance == "existing":
try:
act = syn.getProvenance(ent.id)
except SynapseHTTPError as e:
# Should catch the 404
act = None
elif setProvenance is None or setProvenance.lower() == 'none':
act = None
else:
raise ValueError('setProvenance must be one of None, existing, or traceback')
#Grab file handle createdBy annotation to see the user that created fileHandle
fileHandleList = syn.restGET('/entity/%s/version/%s/filehandles'%(ent.id,ent.versionNumber))
#NOTE: May not always be the first index (need to filter to make sure not PreviewFileHandle)
#Loop through to check which dataFileHandles match and return createdBy
# Look at convenience function
for fileHandle in fileHandleList['list']:
if fileHandle['id'] == ent.dataFileHandleId:
createdBy = fileHandle['createdBy']
break
else:
createdBy = None
#CHECK: If the user created the file, copy the file by using fileHandleId else hard copy
if profile.ownerId == createdBy:
new_ent = File(name=ent.name, parentId=destinationId)
new_ent.dataFileHandleId = ent.dataFileHandleId
else:
#CHECK: If the synapse entity is an external URL, change path and store
if ent.externalURL is None: #and ent.path == None:
#####If you have never downloaded the file before, the path is None
store = True
#This needs to be here, because if the file has never been downloaded before
#there wont be a ent.path
ent = syn.get(entity,downloadFile=store,version=version)
path = ent.path
else:
store = False
ent = syn.get(entity,downloadFile=store,version=version)
path = ent.externalURL
new_ent = File(path, name=ent.name, parentId=destinationId, synapseStore=store)
#Set annotations here
new_ent.annotations = ent.annotations
#Store provenance if act is not None
if act is not None:
new_ent = syn.store(new_ent, activity=act)
else:
new_ent = syn.store(new_ent)
#Leave this return statement for test
return new_ent['id']
if fileName.endswith(('.readsLength.freq', '.txt', '.16_25nt.fq',
'.miRanalyzer.zip')) or re.search(r'_vs_', fileName):
return "analyzed"
return "miRNA"
f['fileFolder'] = f['fileName'].apply(lambda x: getFileFolder(x))
# Get the folder synapse Ids
with open(args.folderInfo) as info:
folderIds = yaml.load(info)
# Link files
for index, row in f.iterrows():
entityFile = File(parentId=folderIds[row['fileFolder']],
name=row['fileName'])
entityFile.annotations = dict(assay=row['assay'],
fileFormat=row['fileFormat'],
cellLine=row['cellLine'],
catalogNumber=row['catalogNumber'],
organ=row['organ'],
diagnosis=row['diagnosis'],
cellType=row['cellType'],
tumorType=row['tumorType'],
consortium="PSON")
newEntity = storePSON(entityFile,
row['filePath'],
contentSize=row['fileSize'],
md5=row['md5'],
syn=syn)
