def def_export_index(name_of_dataset, df_, target_value, number_of_experiment):
list_export = []
X_split = df_.loc[:,~df_.columns.isin([target_])]
y_split = df_.loc[:,df_.columns.isin([target_])]
path_output = tfm_data.def_check_create_path('EXPERIMENTS_kfold', '')
kfold = KFold(n_splits=n_experiment_, shuffle=True)
for train, test in kfold.split(df_):
# list_export.append([tuple(train), tuple(test)])
list_export.append([train, test])
pd.Series(list_export).to_pickle(os.path.join(path_output+str(name_of_dataset)+'.pkl'))
print(os.path.join(path_output+str(name_of_dataset)+'.pkl'))
def def_export_index(name_of_dataset, df_, target_value, list_left_out,
number_of_experiment):
X_split = df_.loc[:, ~df_.columns.isin([target_value])]
y_split = df_.loc[:, df_.columns.isin([target_value])]
path_output = tfm_data.def_check_create_path('EXPERIMENTS', '')
for n_cells_out in list_left_out:
print('n_cells_out,', n_cells_out)
list_export = def_get_n_psplits(X_split, y_split, y_split['cell_type'],
n_cells_out, number_of_experiment)
pd.Series(list_export).to_pickle(
os.path.join(path_output + str(name_of_dataset) + '_cell_out_' +
str(n_cells_out) + '.pkl'))
print(
os.path.join(path_output + str(name_of_dataset) + '_cell_out_' +
str(n_cells_out) + '.pkl'))
config.gpu_options.per_process_gpu_memory_fraction = 0.45
config.gpu_options.allow_growth = True
session = InteractiveSession(config=config)
time_start = dt.datetime.now().time().strftime('%H:%M:%S') # = time.time()
# DEFAULT VALUES for PAPER DESIGN
epochs_default = 100
batch_size_default = 10
# TARGET VARIABLE NAME
target_ = 'cell_type'
TYPE_OF_SCALING = False
# THE LOCATION of THE RESULT of SCORE and MODEL
path_hyperband = tfm_data.def_check_create_path('kt_result',
'DELETE_hyperband')
path_output_result = tfm_data.def_check_create_path(
'kt_result', 'design_' + str(TYPE_OF_SCALING))
path_model = tfm_data.def_check_create_path('kt_result',
'models_' + str(TYPE_OF_SCALING))
# IMPORT EXPERIMENT INDEX
path_experiments = os.path.join(
os.path.dirname(os.getcwd()) + '/data/EXPERIMENTS/')
list_all_model = sorted(os.listdir(path_experiments))
list_experiments = [
string for string in list_all_model
if re.match(re.compile('metabolic_signaling_cell_out_'), string)
]
# LOADING REQUIRED DATASETS
time_start = dt.datetime.now().time().strftime('%H:%M:%S') # = time.time()
# DEFAULT VALUES for PAPER DESIGN
epochs_default = 100
batch_size_default = 10
# TARGET VARIABLE NAME
target_ = 'cell_type'
TYPE_OF_SCALING = [False, True]
for i_row_scaling in TYPE_OF_SCALING:
TYPE_OF_EXPERIMENT = 'no_co_' + str(i_row_scaling)
# THE LOCATION of THE RESULT of SCORE and MODEL
path_model = tfm_data.def_check_create_path('NN_result',
'models_' + TYPE_OF_EXPERIMENT)
# LOADING REQUIRED DATASETS
df_weight_signaling, df_weight_metabolic_signaling = tfm_data.def_load_weight_pathways(
)
df_paper, df_signaling, df_metabolic_signaling = tfm_data.def_load_dataset(
['cell_type'] + list(df_weight_signaling.index.values),
['cell_type'] + list(df_weight_metabolic_signaling.index.values),
row_scaling=i_row_scaling)
df_weight_ppi_tf_signaling, df_weight_ppi_tf_metabolic_signaling = tfm_data.def_load_weight_ppi_tf(
list(df_weight_signaling.index.values),
list(df_weight_metabolic_signaling.index.values))
df_weight_both = pd.concat(
[df_weight_ppi_tf_metabolic_signaling, df_weight_metabolic_signaling],
axis=1)
time_start = dt.datetime.now().time().strftime('%H:%M:%S') # = time.time()
# DEFAULT VALUES for PAPER DESIGN
epochs_default=100
batch_size_default=10
dense_layer=100
# TARGET VARIABLE NAME
target_='cell_type'
TYPE_OF_SCALING = [False, True]
for i_row_scaling in TYPE_OF_SCALING:
# THE LOCATION of THE RESULT of SCORE and MODEL
path_output = tfm_data.def_check_create_path('NN_result_2', 'design_'+str(i_row_scaling))
path_model = tfm_data.def_check_create_path('NN_result_2', 'models_'+str(i_row_scaling))
# IMPORT EXPERIMENT INDEX
path_experiments = os.path.join(os.path.dirname(os.getcwd())+'/data/EXPERIMENTS/')
list_all_model = sorted(os.listdir(path_experiments))
list_experiments = [string for string in list_all_model if re.match(re.compile('default_cell_out'), string)]
# LOADING REQUIRED DATASETS
df_weight_signaling, df_weight_metabolic_signaling = tfm_data.def_load_weight_pathways()
df_paper, df_signaling, df_metabolic_signaling = tfm_data.def_load_dataset(['cell_type']+list(df_weight_signaling.index.values)
, ['cell_type']+list(df_weight_metabolic_signaling.index.values)
, row_scaling=True)
del(df_signaling)
del(df_metabolic_signaling)
time_start = dt.datetime.now().time().strftime('%H:%M:%S') # = time.time()
# DEFAULT VALUES for PAPER DESIGN
epochs_default = 100
batch_size_default = 10
dense_layer = 100
# TARGET VARIABLE NAME
target_ = 'cell_type'
TYPE_OF_SCALING = [False, True]
for i_scaling in TYPE_OF_SCALING:
# THE LOCATION of THE RESULT of SCORE and MODEL
# path_hyperband_ = tfm_data.def_check_create_path('kt_result', '')
path_hyperband_ = tfm_data.def_check_create_path('kt_result', 'delete')
path_output_result = tfm_data.def_check_create_path(
'kt_result', 'design_no_co_' + str(i_scaling))
path_model = tfm_data.def_check_create_path(
'kt_result', 'models_no_co_' + str(i_scaling))
# LOADING REQUIRED DATASETS
df_weight_signaling, df_weight_metabolic_signaling = tfm_data.def_load_weight_pathways(
)
df_paper, df_signaling, df_metabolic_signaling = tfm_data.def_load_dataset(
['cell_type'] + list(df_weight_signaling.index.values),
['cell_type'] + list(df_weight_metabolic_signaling.index.values),
row_scaling=i_scaling,
retrieval=False)
df_weight_dense = pd.DataFrame(df_paper.columns[1:]).set_index('Sample')
from sklearn.preprocessing import MinMaxScaler, StandardScaler
from tensorflow import keras
import tensorflow.keras.backend as K
import warnings
warnings.filterwarnings('ignore')
time_start = dt.datetime.now().time().strftime('%H:%M:%S') # = time.time()
# Clustering is designed with 20 experiments.
# Testing size is %20 (it means that Kfold split is 5)
# and I am applying 4 times to reach the 20 experiments.
# THE LOCATION of THE RESULT
path_output = tfm_data.def_check_create_path(main_folder='clustering_result',
sub_folder='')
# Loading required data
df_weight_pathway_signaling, df_weight_pathway_metabolic_signaling = tfm_data.def_load_weight_pathways(
)
df_paper, df_paper_signaling, df_paper_metabolic_signaling = tfm_data.def_load_dataset(
['cell_type'] + list(df_weight_pathway_signaling.index.values),
['cell_type'] + list(df_weight_pathway_metabolic_signaling.index.values),
row_scaling=False,
retrieval=False)
print('Normalization paper data')
df_scaler_ss = tfm_data.def_dataframe_normalize(df_paper, StandardScaler(),
'cell_type')
print('Normalization signaling data')
df_scaler_ss_signaling = tfm_data.def_dataframe_normalize(