class SC1BSubmissions(SubmissionTools, ST2):
def __init__(self, client=None, name="SC1B"):
super(SC1BSubmissions, self).__init__(client=client, name=name)
from hpn import HPNAdmin
self.hpn = HPNAdmin(client=self.client)
def load_submissions(self, startweek=0, endweek=9, keep_latest=True):
"""Loads all SCORED submissions from SC1B
Attaches the week, status, ranking and zscores
"""
# load all scored submissions
self.submissions = self.hpn.get_submissions_network_insilico(status="SCORED")
print("Got %s SCORED submissions" % len(self.submissions))
# attach week and filter the submissions
self.submissions = self.attach_week_to_submissions(self.submissions, "sc1b")
self.submissions = [sub for sub in self.submissions if sub['week']<=endweek]
self.submissions = [sub for sub in self.submissions if sub['week']>=startweek]
print("Keeping %s submissions in the week range requested" % len(self.submissions))
if keep_latest:
self.submissions = self._keep_latest_only()
print("Keep %s latest scored submissions " % len(self.submissions))
# attach the zscore
print("attaching submissions")
self.submissions = self.attach_status_to_submissions(self.submissions)
print("remove soem users")
self.remove_users()
print("attaching scores and compute final ranking")
self._attach_zscores() # attach zscores
print("all submissions available in the **submissions** attribute")
#return self.submissions
def remove_users(self, userIds=["375805", "1971259", "1961142", "2208193",
"2154231", "2023612", "2200202"]):
"""
* 375805 alphabeta is a test from TC
* 1971259 HD_systems has 2 ID see SC1A remove_users function.
# From Steven's analysis, some results are highly correlated:
* ams1012,cas3,cas4 are the same submission. We keep only ams1012
cas3 1961142, cas4 2208193 should be removed.
* gucas (2154231) is same as Zhangroup. gucas removed
* remove chilin 2023612
* Dream5607 (2200202) and Pitt.transmed are the same. Remove Dream5607
* sfntt has same alias but 2 different userId. Remove 2197351 so that
the latest remains only.
"""
submissions = [x for x in self.submissions if x['userId'] not in userIds]
self.submissions = submissions
def _attach_zscores(self):
"""attach mean zscore, 32 individual zscores
"""
aucs = []
for i, submission in enumerate(self.submissions):
report = json.loads(submission['substatus']['report'])
auc = report['auc']
aucs.append(auc)
ranks = np.argsort(aucs)[::-1] # sort a
for i, rank in enumerate(ranks):
self.submissions[rank]['ranking'] = i
report = json.loads(self.submissions[i]['substatus']['report'])
self.submissions[i]['auc'] = report['auc']
self.submissions[i]['zscore'] = report['score']
def summary_final(self):
header = ("Final rank", "Team name", "userID", "synapse ID", "entityID" , "AUC", "zscore", "p-value")
print("| %12s | %20s | %20s | %20s | %12s | %12s | %12s | %12s |" % header)
print("|%s|%s|%s|%s|%s|%s|" % ("-"*12, "-"*20, "-"*12, "-"*12,"-"*12, "-"*12))
ranks = np.argsort([sub['ranking'] for sub in self.submissions])
results = {}
for count, i in enumerate(ranks):
sub = self.submissions[i]
pvalue = 0
if sub['submitterAlias'] == "ChaosLab":
sub['submitterAlias'] = "FreiburgBiossX"
data = (count+1, sub['submitterAlias'], sub['userId'],
sub['substatus']['id'], sub['substatus']['entityId'],
sub['auc'], sub['zscore'], pvalue)
print("|%12s | %20s | %20s | %20s | %10.6s| %10.6s |%10.6s |%12.6s|" % data)
results[sub['submitterAlias']] = data[:]
return results
