def _get_vencode(self,
amount,
sample_type,
parsed,
thresholds,
n_samples=10000,
using=None):
data = self._get_data(sample_type, parsed, thresholds)
if self.algorithm == "sampling":
vencodes = internals.Vencodes(data,
algorithm="sampling",
number_of_re=self.k,
n_samples=n_samples,
using=using)
elif self.algorithm == "heuristic":
vencodes = internals.Vencodes(data,
algorithm="heuristic",
number_of_re=self.k,
stop=3)
else:
raise AttributeError("Algorithm '{}' not recognized".format(
self.algorithm))
vencodes.next(amount=amount)
if vencodes.vencodes:
return vencodes
else:
raise exceptions.NoVencodeError("No VEnCodes found for {}!".format(
self.cell_type))
def setUpClass(cls):
cls.data = internals.DataTpm(file="parsed", sample_types="primary cells", data_type="promoters", nrows=20000)
cls.data.make_data_celltype_specific("Hepatocyte")
cls.data.filter_by_target_celltype_activity(threshold=1)
cls.data.filter_by_reg_element_sparseness(threshold=0)
cls.data.define_non_target_celltypes_inactivity(threshold=0)
cls.data.sort_sparseness()
cls.vencodes = internals.Vencodes(cls.data, algorithm="heuristic", number_of_re=4, stop=3)
cls.vencodes.next(amount=2)
for celltype in tqdm(setup.re_list, desc="Completed: "):
data.make_data_celltype_specific(celltype)
data_copy = data.copy()
for k in [1, 2]:
results_celltype = []
for n in range(50):
choice = random.sample(range(3), k=k) # chooses a random int from 0 to 2, to later choose a donor.
data.filter_by_target_celltype_activity(threshold=setup.target_celltype_activity, donors=choice)
data.filter_by_reg_element_sparseness(threshold=setup.reg_element_sparseness)
data.define_non_target_celltypes_inactivity(threshold=setup.non_target_celltypes_inactivity)
if setup.algorithm != "sampling":
data.sort_sparseness()
if setup.algorithm == "sampling":
vencodes = internals.Vencodes(data, algorithm="sampling", number_of_re=setup.ven_size, n_samples=10000)
elif setup.algorithm == "heuristic":
vencodes = internals.Vencodes(data, algorithm="heuristic", number_of_re=setup.ven_size, stop=3)
else:
raise AttributeError("Algorithm '{}' not recognized".format(setup.algorithm))
vencodes.next(amount=1)
if vencodes.vencodes:
donors_vencode_data = vencodes.celltype_donors_data.loc[vencodes.vencodes[0]]
assess_if_not_vencode = np.any(donors_vencode_data == 0, axis=0)
result = any(assess_if_not_vencode)
results_celltype.append(not result)
else:
results_celltype.append("")
data = data_copy
results_final[celltype + str(k)] = results_celltype
data_second.make_data_celltype_specific(celltype)
data_second.filter_by_target_celltype_activity(
threshold=setup.second_target_celltype_activity)
data_second.filter_by_reg_element_sparseness(
threshold=setup.second_reg_element_sparseness)
data_second.define_non_target_celltypes_inactivity(
threshold=setup.second_non_target_celltypes_inactivity)
data_second.sort_sparseness()
# Deal with possible dictionaries in celltype list:
if isinstance(celltype, dict):
celltype = list(celltype.keys())[0]
# Launch VEnCode search:
vencodes = internals.Vencodes(data,
algorithm="heuristic",
number_of_re=setup.ven_size,
stop=3)
vencodes.next_heuristic2_vencode(data_second, amount=setup.ven_number)
# Determine e-values:
if vencodes.vencodes:
vencodes.determine_e_values()
e_values = list(vencodes.e_values.values())
if len(e_values) != setup.ven_number:
for i in range(len(e_values), setup.ven_number):
e_values.append("")
results[celltype] = e_values
else:
results[celltype] = [""] * 20
# create a directory to store results
threshold=setup.second_target_celltype_activity)
data_second.filter_by_reg_element_sparseness(
threshold=setup.second_reg_element_sparseness)
data_second.define_non_target_celltypes_inactivity(
threshold=setup.second_non_target_celltypes_inactivity)
data_second.sort_sparseness()
# Deal with possible dictionaries in celltype list:
if isinstance(celltype, dict):
celltype = list(celltype.keys())[0]
data_second_original = data_second.copy()
results[celltype] = []
for k in range(1, 5):
# Launch VEnCode search:
vencodes = internals.Vencodes(data,
algorithm="heuristic",
number_of_re=k,
stop=3)
vencodes.next_heuristic2_vencode(data_second, amount=setup.ven_number)
# Determine e-values:
if vencodes.vencodes:
vencodes.determine_e_values()
best = []
e_values = vencodes.e_values.copy()
for i in range(5):
try:
best_partial = max(e_values, key=lambda key: e_values[key])
except ValueError:
break
best.append(best_partial)
try:
def setUp(self):
self.vencodes = internals.Vencodes(self.data, algorithm="heuristic", number_of_re=4)
def setUp(self):
self.vencodes = internals.Vencodes(self.data, algorithm="sampling", number_of_re=4, n_samples=10000)