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)
