def get_sig(run_id, sig_level = .05):
"""
Returns the significant networks at the given significance by
Benjamini-Hochberg
"""
myitem = None
for item in DataForDisplay.query(run_id):
if not myitem:
myitem = item
elif myitem and item.timestamp > myitem.timestamp:
myitem = item
bucket = myitem.data_bucket
pv_file = myitem.data_file
conn = boto.connect_s3()
b = conn.get_bucket(bucket)
k = b.get_key(pv_file)
with tempfile.TemporaryFile() as fp:
k.get_contents_to_file(fp)
fp.seek(0)
table = pandas.read_csv(fp, sep='\t')
nv = NetworkTSV()
cutoffs = nv.get_fdr_cutoffs( myitem.identifier, myitem.timestamp, [sig_level] )
valid = []
for k,v in cutoffs.iteritems():
for cut in v.itervalues():
valid += table[table[k] <= cut]['networks'].tolist()
return list(set(valid))
def get_fdr_cutoffs( self, identifier, timestamp, alphas=[.05]):
"""
By benjamini-hochberg
"""
res = DataForDisplay.get(identifier, timestamp)
s3 = boto.connect_s3()
bucket = s3.get_bucket( res.data_bucket )
k = bucket.get_key( res.data_file )
with tempfile.TemporaryFile() as fp:
k.get_contents_to_file(fp)
fp.seek(0)
res = stat.get_fdr_cutoffs(fp, alphas=alphas)
return res
def get_nets_for_display():
from datadirac.aggregate import DataForDisplay
import json
res = DataForDisplay.scan()
rlist = [r.attribute_values for r in res]
for a in rlist:
for k,v in a.iteritems():
if type(v) is set:
a[k]=list(v)
if res:
return Response( json.dumps(rlist),
mimetype='application/json')
else:
abort(400)
def set_qval_table( self, identifier, timestamp ):
res = DataForDisplay.get(identifier, timestamp)
s3 = boto.connect_s3()
bucket = s3.get_bucket( res.data_bucket )
k = bucket.get_key( res.data_file )
with tempfile.TemporaryFile() as fp:
k.get_contents_to_file(fp)
fp.seek(0)
qv = stat.get_qval_table(fp)
with tempfile.TemporaryFile() as fp2:
qv.to_csv( fp2, sep='\t', index_label='networks' )
fp2.seek(0)
k = Key(bucket)
k.key = 'qvals-' + identifier + '-' + timestamp + '.tsv'
k.set_contents_from_file(fp2)
res.qvalue_file = 'qvals-' + identifier + '-' + timestamp + '.tsv'
res.save()
def _available(self):
res = DataForDisplay.scan()
results = {}
for r in res:
results[r.identifier +'-' +r.timestamp] = r.attribute_values
return results