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)
'tissueType':
['Frontal Pole', 'Superior Temporal Gyrus', 'Parahippocampal Gyrus'],
'tissueTypeAbrv': ['FP', 'STG', 'PHG'],
'name':
'AMP-AD_MSBB_MSSM_IlluminaHiSeq2500_mRNA_normalized-sex-race-age-RIN-PMI-batch-site.corrected.csv'
},
}
for id, v in toMove.items():
ent = syn.get(id)
print v['name']
os.rename(ent.path, v['name'])
f = File(v['name'], parentId=v['parentId'], name=v['name'][7:-4])
print f.name
f.consortium, f.study, f.center, f.disease = consortium, study, center, disease
f.dataType = v['dataType']
f.platfrom = v['platform']
if 'tissueTypeAbrv' in v:
f.tissueTypeAbrv = v['tissueTypeAbrv']
f.tissueType = v['tissueType']
f.fileType = fileType
f.organism = organism
f = syn.store(
f,
used=[id],
executed=[
'https://github.com/Sage-Bionetworks/ampAdScripts/blob/e71bbde262625e6999ea9defd98e10fce8f3c542/Mount-Sinai/migrateMSBBMetaAndRNASeq.py'
],
activityName='Data migration')
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')
'tissueTypeAbrv': ['FP', 'STG', 'PHG'],
'name' :'AMP-AD_MSBB_MSSM_IlluminaHiSeq2500_mRNA_rawCounts.tsv'},
'syn2920161':{'parentId' :'syn3157743', #'normalized.sex_race_age_RIN_PMI_batch_site.corrected.csv'
'dataType': 'mRNA',
'platform': 'IlluminaHiSeq2500',
'tissueType':['Frontal Pole', 'Superior Temporal Gyrus','Parahippocampal Gyrus'],
'tissueTypeAbrv': ['FP', 'STG', 'PHG'],
'name' :'AMP-AD_MSBB_MSSM_IlluminaHiSeq2500_mRNA_normalized-sex-race-age-RIN-PMI-batch-site.corrected.csv'},
}
for id, v in toMove.items():
ent = syn.get(id)
print v['name']
os.rename(ent.path, v['name'])
f = File(v['name'], parentId=v['parentId'], name=v['name'][7:-4])
print f.name
f.consortium, f.study, f.center, f.disease = consortium, study, center, disease
f.dataType = v['dataType']
f.platfrom = v['platform']
if 'tissueTypeAbrv' in v:
f.tissueTypeAbrv = v['tissueTypeAbrv']
f.tissueType = v['tissueType']
f.fileType = fileType
f.organism = organism
f = syn.store(f, used = [id], executed=['https://github.com/Sage-Bionetworks/ampAdScripts/blob/e71bbde262625e6999ea9defd98e10fce8f3c542/Mount-Sinai/migrateMSBBMetaAndRNASeq.py'],
activityName='Data migration')