def storeFile(syn,
fileName,
parentId,
center,
fileFormat,
dataSubType,
platform=None,
cBioFileFormat=None,
used=None):
logger.info("STORING FILES")
fileEnt = File(fileName, parent=parentId)
fileEnt.center = center
fileEnt.species = "Human"
fileEnt.consortium = 'GENIE'
fileEnt.dataType = "genomicVariants"
fileEnt.fundingAgency = "AACR"
fileEnt.assay = 'targetGeneSeq'
fileEnt.fileFormat = fileFormat
fileEnt.dataSubType = dataSubType
fileEnt.fileStage = "staging"
fileEnt.platform = platform
if platform is not None:
fileEnt.platform = platform
if cBioFileFormat is not None:
fileEnt.cBioFileFormat = cBioFileFormat
ent = syn.store(fileEnt, used=used)
return (ent)
def uploadToSynapse(f):
"""Given a filepath extracts metadata and uploads to Synapse"""
center, sample_id, workflow_name, date, call_type, dataType, fileType = ['']*7
url = URLBASE+f
if 'OICR_BL' in f: center = 'oicr_bl'
elif 'CRG/clindel/somatic' in f: center = 'crg_clindel'
else: center = f.split('/')[4]
filename = f.split('/')[-1]
if center in ('yale', 'wustl', 'LOHcomplete'):
if filename =='bd829214-f230-4331-b234-def10bbe7938CNV.vcf.gz':
sample_id, dataType, fileType='bd829214-f230-4331-b234-def10bbe7938', 'cnv', 'vcf'
else:
sample_id, dataType = filename.lower().split('.')[:2]
fileType = [i for i in filename.split('.')[2:] if i != 'gz'][-1]
elif center in ('broad', 'BSC', 'oicr_sga', 'mda_kchen', 'MDA_HGSC', 'mcgill_popsv', 'sfu', 'UCSC', 'oicr_bl', 'Synteka_pgm21', 'crg_clindel'):
sample_id, workflow_name, date, call_type, dataType = filename.replace('indels', 'indel', split('.')[:5])
fileType = [i for i in filename.split('.')[5:] if i != 'gz'][-1]
else:
print 'Not uploading:', f
return
print center, workflow_name, date, call_type, dataType, fileType
file = File(url, parentId=DIRS[center], synapseStore=False)
file.center = center.lower()
file.sample_id = sample_id
file.workflow_name = workflow_name
file.date = date
file.call_type = call_type
file.dataType = 'DNA'
file.disease = 'Cancer'
file.dataSubType = dataType
file.fileType = fileType
#file.analysis_id_tumor = ?????
syn.store(file, forceVersion=False)
def storeFile(fileName, stagingID, used, center, annotations, meta=False):
print("STORING FILES")
fileEnt = File(fileName, parent = stagingID)
fileEnt.center = center
fileEnt.dataSubType = annotations.get("dataSubType",'')
fileEnt.dataType = annotations.get("dataType",'')
fileEnt.disease = 'cancer'
fileEnt.fileType = annotations.get("fileType",'')
fileEnt.organism = 'H**o Sapiens'
fileEnt.platform = annotations.get("platform",'')
fileEnt.tissueSource = annotations.get("tissueSource",'')
fileEnt.consortium = 'GENIE'
if meta:
fileEnt.fileType = "txt"
fileEnt.dataType = "meta"
fileEnt.fileStage = "staging"
ent = syn.store(fileEnt,annotations = used)
return(ent)
PLATFORM_MAP = {'133AB': 'AffymetrixU133AB',
'Plus2': 'AffymetrixU133Plus2'}
query = 'select id, name from entity where parentId=="%s"' %OLDPARENTID
df = synapseHelpers.query2df(syn.chunkedQuery(query))
for i in range(1,df.shape[0]):
row = df.ix[i, :]
ent = syn.get(row.id)
fStudy, fTissue, fPlatform, fDatatype, fRest = ent.name.split('_')
name = 'AMP-AD_MSBB_MSSM_%s_%s_%s' % (PLATFORM_MAP[fPlatform],
TISSUEABRMAP[fTissue][0], fRest)
print name
os.rename(ent.path, name)
f = File(name, parentId=NEWPARENTID, name=name[7:])
f.consortium = 'AMP-AD'
f.study = 'MSBB'
f.center = 'MSSM'
f.dataType = 'mRNA'
f.disease = 'Alzheimers Disease'
f.platfrom = PLATFORM_MAP[fPlatform]
f.tissueTypeAbrv = TISSUEABRMAP[fTissue][1]
f.tissueType = TISSUEABRMAP[fTissue][0]
f.dataSubType = 'geneExp'
f.fileType = 'genomicMatrix'
f.organism = 'human'
f = syn.store(f, used = [ent], executed=['https://github.com/Sage-Bionetworks/ampAdScripts/blob/4d7d6b78b1e73058483354a1a18bff7422966a4b/Mount-Sinai/migrateMSBBExpression.py'], activityName='Data migration')
ent = syn.get(row.id)
# fStudy, fTissue, fPlatform, fDatatype, fProcess, fRest = ent.name.split('_')
# name = 'AMP-AD_MSBB_MSSM_%s_%s_%s' % (PLATFORM_MAP[fPlatform],
# TISSUEABRMAP[fTissue][0], fRest
fStudy, fTissue, fPlatform, fDatatype, fProcess, fRest = ent.name.split('_')
name = 'AMP-AD_MSBB_MSSM_%s_%s_%s_%s' % (PLATFORM_MAP[fPlatform],
TISSUEABRMAP[fTissue][0],
fProcess,
fRest)
print name
used = syn.query('select id from file where parentId=="%s" and tissueType=="%s" ' %(EXPRFILE_PARENTID, TISSUEABRMAP[fTissue][0]))['results'][0].values()[0]
os.rename(ent.path, name)
f = File(name, parentId=NEWPARENTID, name=name[7:])
f.consortium = 'AMP-AD'
f.study = 'MSBB'
f.center = 'MSSM'
f.dataType = 'mRNA'
f.disease = 'Alzheimers Disease'
f.platfrom = PLATFORM_MAP[fPlatform]
f.tissueTypeAbrv = TISSUEABRMAP[fTissue][1]
f.tissueType = TISSUEABRMAP[fTissue][0]
f.dataSubType = 'CoExpression'
f.fileType = 'genomicMatrix'
f.organism = 'human'
f = syn.store(f, used = [used], executed=['syn2731322'],
activityName='Weighted coexpression network modules using Coexpp v 0.1.0')
# name = 'AMP-AD_MSBB_MSSM_%s_%s_%s' % (PLATFORM_MAP[fPlatform],
# TISSUEABRMAP[fTissue][0], fRest
fStudy, fTissue, fPlatform, fDatatype, fProcess, fRest = ent.name.split(
'_')
name = 'AMP-AD_MSBB_MSSM_%s_%s_%s_%s' % (
PLATFORM_MAP[fPlatform], TISSUEABRMAP[fTissue][0], fProcess, fRest)
print name
used = syn.query(
'select id from file where parentId=="%s" and tissueType=="%s" ' %
(EXPRFILE_PARENTID,
TISSUEABRMAP[fTissue][0]))['results'][0].values()[0]
os.rename(ent.path, name)
f = File(name, parentId=NEWPARENTID, name=name[7:])
f.consortium = 'AMP-AD'
f.study = 'MSBB'
f.center = 'MSSM'
f.dataType = 'mRNA'
f.disease = 'Alzheimers Disease'
f.platfrom = PLATFORM_MAP[fPlatform]
f.tissueTypeAbrv = TISSUEABRMAP[fTissue][1]
f.tissueType = TISSUEABRMAP[fTissue][0]
f.dataSubType = 'CoExpression'
f.fileType = 'genomicMatrix'
f.organism = 'human'
f = syn.store(f,
used=[used],
executed=['syn2731322'],
activityName=
'Weighted coexpression network modules using Coexpp v 0.1.0')