def clone_cohort(request, cohort_id):
if debug: print >> sys.stderr,'Called '+sys._getframe().f_code.co_name
redirect_url = 'cohort_details'
parent_cohort = Cohort.objects.get(id=cohort_id)
new_name = 'Copy of %s' % parent_cohort.name
cohort = Cohort.objects.create(name=new_name)
cohort.save()
# If there are sample ids
samples = Samples.objects.filter(cohort=parent_cohort).values_list('sample_id', 'study_id')
sample_list = []
for sample in samples:
sample_list.append(Samples(cohort=cohort, sample_id=sample[0], study_id=sample[1]))
Samples.objects.bulk_create(sample_list)
# TODO Some cohorts won't have them at the moment. That isn't a big deal in this function
# If there are patient ids
patients = Patients.objects.filter(cohort=parent_cohort).values_list('patient_id', flat=True)
patient_list = []
for patient_code in patients:
patient_list.append(Patients(cohort=cohort, patient_id=patient_code))
Patients.objects.bulk_create(patient_list)
# Clone the filters
filters = Filters.objects.filter(resulting_cohort=parent_cohort).values_list('name', 'value')
# ...but only if there are any (there may not be)
if filters.__len__() > 0:
filters_list = []
for filter_pair in filters:
filters_list.append(Filters(name=filter_pair[0], value=filter_pair[1], resulting_cohort=cohort))
Filters.objects.bulk_create(filters_list)
# Set source
source = Source(parent=parent_cohort, cohort=cohort, type=Source.CLONE)
source.save()
# Set permissions
perm = Cohort_Perms(cohort=cohort, user=request.user, perm=Cohort_Perms.OWNER)
perm.save()
# Store cohort to BigQuery
project_id = settings.BQ_PROJECT_ID
cohort_settings = settings.GET_BQ_COHORT_SETTINGS()
bcs = BigQueryCohortSupport(project_id, cohort_settings.dataset_id, cohort_settings.table_id)
bcs.add_cohort_with_sample_barcodes(cohort.id, samples)
return redirect(reverse(redirect_url,args=[cohort.id]))
def set_operation(request):
if debug: print >> sys.stderr,'Called '+sys._getframe().f_code.co_name
redirect_url = '/cohorts/'
if request.POST:
name = request.POST.get('name').encode('utf8')
cohorts = []
base_cohort = None
subtract_cohorts = []
notes = ''
patients = []
samples = []
op = request.POST.get('operation')
if op == 'union':
notes = 'Union of '
cohort_ids = request.POST.getlist('selected-ids')
cohorts = Cohort.objects.filter(id__in=cohort_ids, active=True, cohort_perms__in=request.user.cohort_perms_set.all())
first = True
ids = ()
for cohort in cohorts:
if first:
notes += cohort.name
first = False
else:
notes += ', ' + cohort.name
ids += (cohort.id,)
patients = Patients.objects.filter(cohort_id__in=ids).distinct().values_list('patient_id', flat=True)
samples = Samples.objects.filter(cohort_id__in=ids).distinct().values_list('sample_id', 'study_id')
elif op == 'intersect':
cohort_ids = request.POST.getlist('selected-ids')
cohorts = Cohort.objects.filter(id__in=cohort_ids, active=True, cohort_perms__in=request.user.cohort_perms_set.all())
request.user.cohort_perms_set.all()
if len(cohorts):
cohort_patients = set(Patients.objects.filter(cohort=cohorts[0]).values_list('patient_id'))
cohort_samples = set(Samples.objects.filter(cohort=cohorts[0]).values_list('sample_id', 'study_id'))
notes = 'Intersection of ' + cohorts[0].name
# print "Start of intersection with %s has %d" % (cohorts[0].name, len(cohort_samples))
for i in range(1, len(cohorts)):
cohort = cohorts[i]
notes += ', ' + cohort.name
cohort_patients = cohort_patients.intersection(Patients.objects.filter(cohort=cohort).values_list('patient_id'))
cohort_samples = cohort_samples.intersection(Samples.objects.filter(cohort=cohort).values_list('sample_id', 'study_id'))
# se1 = set(x[0] for x in s1)
# se2 = set(x[0] for x in s2)
# TODO: work this out with user data when activated
# cohort_samples = cohort_samples.extra(
# tables=[Samples._meta.db_table+"` AS `t"+str(1)], # TODO This is ugly :(
# where=[
# 't'+str(i)+'.sample_id = ' + Samples._meta.db_table + '.sample_id',
# 't'+str(i)+'.study_id = ' + Samples._meta.db_table + '.study_id',
# 't'+str(i)+'.cohort_id = ' + Samples._meta.db_table + '.cohort_id',
# ]
# )
# cohort_patients = cohort_patients.extra(
# tables=[Patients._meta.db_table+"` AS `t"+str(1)], # TODO This is ugly :(
# where=[
# 't'+str(i)+'.patient_id = ' + Patients._meta.db_table + '.patient_id',
# 't'+str(i)+'.cohort_id = ' + Patients._meta.db_table + '.cohort_id',
# ]
# )
patients = list(cohort_patients)
samples = list(cohort_samples)
elif op == 'complement':
base_id = request.POST.get('base-id')
subtract_ids = request.POST.getlist('subtract-ids')
base_patients = Patients.objects.filter(cohort_id=base_id)
subtract_patients = Patients.objects.filter(cohort_id__in=subtract_ids).distinct()
cohort_patients = base_patients.exclude(patient_id__in=subtract_patients.values_list('patient_id', flat=True))
patients = cohort_patients.values_list('patient_id', flat=True)
base_samples = Samples.objects.filter(cohort_id=base_id)
subtract_samples = Samples.objects.filter(cohort_id__in=subtract_ids).distinct()
cohort_samples = base_samples.exclude(sample_id__in=subtract_samples.values_list('sample_id', flat=True))
samples = cohort_samples.values_list('sample_id', 'study_id')
notes = 'Subtracted '
base_cohort = Cohort.objects.get(id=base_id)
subtracted_cohorts = Cohort.objects.filter(id__in=subtract_ids)
first = True
for item in subtracted_cohorts:
if first:
notes += item.name
first = False
else:
notes += ', ' + item.name
notes += ' from %s.' % base_cohort.name
if len(samples) or len(patients):
new_cohort = Cohort.objects.create(name=name)
perm = Cohort_Perms(cohort=new_cohort, user=request.user, perm=Cohort_Perms.OWNER)
perm.save()
# Store cohort to BigQuery
project_id = settings.BQ_PROJECT_ID
cohort_settings = settings.GET_BQ_COHORT_SETTINGS()
bcs = BigQueryCohortSupport(project_id, cohort_settings.dataset_id, cohort_settings.table_id)
bcs.add_cohort_with_sample_barcodes(new_cohort.id, samples)
# Store cohort to CloudSQL
patient_list = []
for patient in patients:
patient_list.append(Patients(cohort=new_cohort, patient_id=patient))
Patients.objects.bulk_create(patient_list)
sample_list = []
for sample in samples:
sample_list.append(Samples(cohort=new_cohort, sample_id=sample[0], study_id=sample[1]))
Samples.objects.bulk_create(sample_list)
# Create Sources
if op == 'union' or op == 'intersect':
for cohort in cohorts:
source = Source.objects.create(parent=cohort, cohort=new_cohort, type=Source.SET_OPS, notes=notes)
source.save()
elif op=='complement':
source = Source.objects.create(parent=base_cohort, cohort=new_cohort, type=Source.SET_OPS, notes=notes)
source.save()
for cohort in subtract_cohorts:
source = Source.objects.create(parent=cohort, cohort=new_cohort, type=Source.SET_OPS, notes=notes)
source.save()
else:
message = 'Operation resulted in empty set of samples and patients. Cohort not created.'
messages.warning(request, message)
return redirect('cohort_list')
return redirect(redirect_url)
def save_cohort(request, workbook_id=None, worksheet_id=None, create_workbook=False):
if debug: print >> sys.stderr,'Called '+sys._getframe().f_code.co_name
redirect_url = reverse('cohort_list')
samples = []
patients = []
name = ''
user_id = request.user.id
parent = None
if request.POST:
name = request.POST.get('name')
source = request.POST.get('source')
deactivate_sources = request.POST.get('deactivate_sources')
filters = request.POST.getlist('filters')
projects = request.user.project_set.all()
token = SocialToken.objects.filter(account__user=request.user, account__provider='Google')[0].token
data_url = METADATA_API + 'v2/metadata_sample_list'
payload = {
'token': token
}
# Given cohort_id is the only source id.
if source:
# Only ever one source
# data_url += '&cohort_id=' + source
payload['cohort_id'] = source
parent = Cohort.objects.get(id=source)
if deactivate_sources:
parent.active = False
parent.save()
if filters:
filter_obj = []
for filter in filters:
tmp = json.loads(filter)
key = tmp['feature']['name']
val = tmp['value']['name']
if 'id' in tmp['feature'] and tmp['feature']['id']:
key = tmp['feature']['id']
if 'id' in tmp['value'] and tmp['value']['id']:
val = tmp['value']['id']
if key == 'user_projects':
proj = projects.get(id=val)
studies = proj.study_set.all()
for study in studies:
filter_obj.append({
'key': 'user_studies',
'value': str(study.id)
})
else :
filter_obj.append({
'key': key,
'value': val
})
if len(filter_obj):
# data_url += '&filters=' + re.sub(r'\s+', '', urllib.quote( json.dumps(filter_obj) ))
payload['filters'] = json.dumps(filter_obj)
result = urlfetch.fetch(data_url, method=urlfetch.POST, payload=json.dumps(payload), deadline=60, headers={'Content-Type': 'application/json'})
items = json.loads(result.content)
#it is possible the the filters are creating a cohort with no samples
if int(items['count']) == 0 :
messages.error(request, 'The filters selected returned 0 samples. Please alter your filters and try again')
redirect_url = reverse('cohort')
else :
items = items['items']
for item in items:
samples.append(item['sample_barcode'])
#patients.append(item['ParticipantBarcode'])
# Create new cohort
cohort = Cohort.objects.create(name=name)
cohort.save()
# If there are sample ids
sample_list = []
for item in items:
study = None
if 'study_id' in item:
study = item['study_id']
sample_list.append(Samples(cohort=cohort, sample_id=item['sample_barcode'], study_id=study))
Samples.objects.bulk_create(sample_list)
# TODO This would be a nice to have if we have a mapped ParticipantBarcode value
# TODO Also this gets weird with mixed mapped and unmapped ParticipantBarcode columns in cohorts
# If there are patient ids
# If we are *not* using user data, get participant barcodes from metadata_data
if not USER_DATA_ON:
participant_url = METADATA_API + ('v2/metadata_participant_list?cohort_id=%s' % (str(cohort.id),))
participant_result = urlfetch.fetch(participant_url, deadline=120)
participant_items = json.loads(participant_result.content)
participant_list = []
for item in participant_items['items']:
participant_list.append(Patients(cohort=cohort, patient_id=item['sample_barcode']))
Patients.objects.bulk_create(participant_list)
# Set permission for user to be owner
perm = Cohort_Perms(cohort=cohort, user=request.user, perm=Cohort_Perms.OWNER)
perm.save()
# Create the source if it was given
if source:
Source.objects.create(parent=parent, cohort=cohort, type=Source.FILTERS).save()
# Create filters applied
if filters:
for filter in filter_obj:
Filters.objects.create(resulting_cohort=cohort, name=filter['key'], value=filter['value']).save()
# Store cohort to BigQuery
project_id = settings.BQ_PROJECT_ID
cohort_settings = settings.GET_BQ_COHORT_SETTINGS()
bcs = BigQueryCohortSupport(project_id, cohort_settings.dataset_id, cohort_settings.table_id)
bcs.add_cohort_with_sample_barcodes(cohort.id, cohort.samples_set.values_list('sample_id','study_id'))
# Check if coming from applying filters and redirect accordingly
if 'apply-filters' in request.POST:
redirect_url = reverse('cohort_details',args=[cohort.id])
messages.info(request, 'Changes applied successfully.')
else:
redirect_url = reverse('cohort_list')
messages.info(request, 'Cohort, %s, created successfully.' % cohort.name)
if workbook_id and worksheet_id :
Worksheet.objects.get(id=worksheet_id).add_cohort(cohort)
redirect_url = reverse('worksheet_display', kwargs={'workbook_id':workbook_id, 'worksheet_id' : worksheet_id})
elif create_workbook :
workbook_model = Workbook.create("default name", "This is a default workbook description", request.user)
worksheet_model = Worksheet.create(workbook_model.id, "worksheet 1","This is a default description")
worksheet_model.add_cohort(cohort)
redirect_url = reverse('worksheet_display', kwargs={'workbook_id':workbook_model.id, 'worksheet_id' : worksheet_model.id})
return redirect(redirect_url) # redirect to search/ with search parameters just saved