def __result_parser(self, result):
'''parse result input example:
[{'num_variants_CFisher': 3, 'pvalue_CFisher': 0.06156133406564546, 'total_mac_CFisher': 4, 'sample_size_CFisher': 400, 'statistic_CFisher': 1.7976931348623157e+308}, {'statistic_WSSRankTest': 1784.0, 'total_mac_WSSRankTest': 4, 'pvalue_WSSRankTest': 0.027761334758504563, 'num_variants_WSSRankTest': 3, 'sample_size_WSSRankTest': 400}]
output format: a dictionary of properly organized power and data statistic.
'''
if len(result) == 0:
raise NullResultException
out = {}
names = []
for item in result:
for k in item:
# record statistics
for entry in ['num_variants', 'total_mac', 'sample_size']:
if k.startswith(entry) and '{}_analyzed'.format(entry) not in out:
out['{}_analyzed'.format(entry)] = item[k]
# record p-values
if k.startswith('pvalue_'):
name = k[7:]
if name in names:
# resolve name conflict
i = 0
while name + ('_{}'.format(i) if i else '') in names:
i += 1
name += '_{}'.format(i)
names.append(name)
out['power_{}'.format(name)] = item[k]
if [is_null(out[k]) for k in out].count(True) == len(out):
raise NullResultException
return out
def __result_parser(self, result):
'''parse result input example:
[{'num_variants_CFisher': 3, 'pvalue_CFisher': 0.06156133406564546, 'total_mac_CFisher': 4, 'sample_size_CFisher': 400, 'statistic_CFisher': 1.7976931348623157e+308}, {'statistic_WSSRankTest': 1784.0, 'total_mac_WSSRankTest': 4, 'pvalue_WSSRankTest': 0.027761334758504563, 'num_variants_WSSRankTest': 3, 'sample_size_WSSRankTest': 400}]
output format: a dictionary of properly organized power and data statistic.
'''
if len(result) == 0:
raise NullResultException
out = {}
names = []
for item in result:
for k in item:
# record statistics
for entry in ['num_variants', 'total_mac', 'sample_size']:
if k.startswith(entry) and '{}_analyzed'.format(
entry) not in out:
out['{}_analyzed'.format(entry)] = item[k]
# record p-values
if k.startswith('pvalue_'):
name = k[7:]
if name in names:
# resolve name conflict
i = 0
while name + ('_{}'.format(i) if i else '') in names:
i += 1
name += '_{}'.format(i)
names.append(name)
out['power_{}'.format(name)] = item[k]
if [is_null(out[k]) for k in out].count(True) == len(out):
raise NullResultException
return out
def preprocess(self, data, exclude=[], hide=[]):
"""format result to output, messy messy"""
multicols = []
for key in list(data.keys()):
if key in ["pool"]:
continue
# collect results from RunningStat
if isinstance(data[key], L.RunningStat):
data[key], data[key + "_median"], data[key + "_std"] = (
data[key].mean(),
data[key].left(),
data[key].sd(),
)
continue
# delete trivial data
if is_null(data[key]):
del data[key]
continue
# adjust key names
if key in hide:
data["_" + key] = data[key]
del data[key]
continue
# multi column information
if type(data[key]) is list:
multicols.append(key)
if len([x for x in data.keys() if x.startswith("power")]):
# manually combine multiple power analysis methods into single column
for key in ["power", "method"]:
data[key] = []
for key in list(data.keys()):
if key.startswith("power") and key not in ["power", "power_std", "power_median"]:
if key.endswith("_std"):
if "power_std" not in data:
data["power_std"] = []
if "default" in key:
data["power_std"].append(data[key])
else:
# adjust standard error
data["power_std"].append(data[key] / np.sqrt(data["_replicates"]))
elif key.endswith("_median"):
if "power_median" not in data:
data["power_median"] = []
data["power_median"].append(data[key])
else:
data["power"].append(data[key])
data["method"].append(re.sub("power_", "", key))
del data[key]
# expand table
for key in list(data.keys()):
if key not in ["power", "power_std", "power_median", "method", "model"] + multicols:
data[key] = [data[key]] * (max(len(data["power"]), 1) if "power" in data else 1)
# manually create ordered column names
colnames = ["title", "name"] + sorted(
[x for x in list(data.keys()) if x not in exclude],
key=lambda x: x.replace("_", "|").replace("method", "AAA").replace("power", "AAB"),
)
# return: data, single row colnames, multi row colnames
return data, [x for x in colnames if x not in multicols], [x for x in colnames if x in multicols]
def preprocess(self, data, exclude = [], hide = []):
'''format result to output, messy messy'''
multicols = []
for key in list(data.keys()):
if key in ['pool']:
continue
# collect results from RunningStat
if isinstance(data[key], L.RunningStat):
data[key], data[key + '_median'], data[key + '_std'] = \
data[key].mean(), data[key].left(), data[key].sd()
continue
# delete trivial data
if is_null(data[key]):
del data[key]
continue
# adjust key names
if key in hide:
data["_" + key] = data[key]
del data[key]
continue
# multi column information
if type(data[key]) is list:
multicols.append(key)
if len([x for x in data.keys() if x.startswith('power')]):
# manually combine multiple power analysis methods into single column
for key in ['power', 'method']:
data[key] = []
for key in list(data.keys()):
if key.startswith('power') and key not in ['power', 'power_std', 'power_median']:
if key.endswith('_std'):
if 'power_std' not in data:
data['power_std'] = []
if 'default' in key:
data['power_std'].append(data[key])
else:
# adjust standard error
data['power_std'].append(data[key]/np.sqrt(data['_replicates']))
elif key.endswith('_median'):
if 'power_median' not in data:
data['power_median'] = []
data['power_median'].append(data[key])
else:
data['power'].append(data[key])
data['method'].append(re.sub('power_', '', key))
del data[key]
# expand table
for key in list(data.keys()):
if key not in ['power', 'power_std', 'power_median', 'method', 'model'] + multicols:
data[key] = [data[key]] * (max(len(data['power']), 1) if 'power' in data else 1)
# manually create ordered column names
colnames = ['title', 'name'] + sorted([x for x in list(data.keys()) if x not in exclude],
key = lambda x: x.replace("_", "|").replace('method', 'AAA').replace('power', 'AAB'))
# return: data, single row colnames, multi row colnames
return data, [x for x in colnames if x not in multicols], [x for x in colnames if x in multicols]
def __get_type(self, values):
types = []
for item in values:
# Here assume the input list has the same type on each element
# Which is true in simulation data
if isinstance(item, list):
for i in item:
if not is_null(i):
item = i
break
try:
item = float(item)
types.append('number')
except:
types.append('string')
return types
def __get_type(self, values):
types = []
for item in values:
# Here assume the input list has the same type on each element
# Which is true in simulation data
if isinstance(item, list):
for i in item:
if not is_null(i):
item = i
break
try:
item = float(item)
types.append('number')
except:
types.append('string')
return types
def preprocess(self, data, exclude=[], hide=[]):
'''format result to output, messy messy'''
multicols = []
for key in list(data.keys()):
if key in ['pool']:
continue
# collect results from RunningStat
if isinstance(data[key], L.RunningStat):
data[key], data[key + '_median'], data[key + '_std'] = \
data[key].mean(), data[key].left(), data[key].sd()
continue
# delete trivial data
if is_null(data[key]):
del data[key]
continue
# adjust key names
if key in hide:
data["_" + key] = data[key]
del data[key]
continue
# multi column information
if type(data[key]) is list:
multicols.append(key)
if len([x for x in data.keys() if x.startswith('power')]):
# manually combine multiple power analysis methods into single column
for key in ['power', 'method']:
data[key] = []
for key in list(data.keys()):
if key.startswith('power') and key not in [
'power', 'power_std', 'power_median'
]:
if key.endswith('_std'):
if 'power_std' not in data:
data['power_std'] = []
if 'default' in key:
data['power_std'].append(data[key])
else:
# adjust standard error
data['power_std'].append(
data[key] / np.sqrt(data['_replicates']))
elif key.endswith('_median'):
if 'power_median' not in data:
data['power_median'] = []
data['power_median'].append(data[key])
else:
data['power'].append(data[key])
data['method'].append(re.sub('power_', '', key))
del data[key]
# expand table
for key in list(data.keys()):
if key not in [
'power', 'power_std', 'power_median', 'method', 'model'
] + multicols:
data[key] = [data[key]] * (max(len(data['power']), 1)
if 'power' in data else 1)
# manually create ordered column names
colnames = ['title', 'name'] + sorted(
[x for x in list(data.keys()) if x not in exclude],
key=lambda x: x.replace("_", "|").replace('method', 'AAA').replace(
'power', 'AAB'))
# return: data, single row colnames, multi row colnames
return data, [x for x in colnames if x not in multicols
], [x for x in colnames if x in multicols]
