def load_sample_data(path,
samples_selected=['sample1'],
batch_names=['batch1', 'batch2']):
"""
Function to load and merge data for samples and convert it to a desired format
"""
df_full = None
for sample in samples_selected:
x1_train, x1_test, x2_train, x2_test = load_data_basic(
path, sample=sample, batch_names=batch_names, panel=None)
tmp_ = pd.concat([x1_train, x2_train])
if df_full is None:
df_full = tmp_
else:
df_full = pd.concat([df_full, tmp_], axis=0)
metadata_batch = [i.split('_')[0] for i in df_full.index]
metadata_cell = [i.split('_')[-1] for i in df_full.index]
metadata_sample = [i.split('_')[1] for i in df_full.index]
df_full['metadata_batch'] = metadata_batch
df_full['metadata_celltype'] = metadata_cell
df_full['metadata_sample'] = metadata_sample
df_full = df_full.dropna(axis=1)
df_full = df_full.reset_index(drop=True)
return (df_full)
# If at save interval => save generated image samples
if epoch % sample_interval == 0:
print('generating plots and saving outputs')
gx1 = self.generator.predict(x1_train_df)
self.generator.save(os.path.join('models_' + self.modelname, fname, 'generator' + str(epoch)))
save_info.save_dataframes(epoch, x1_train_df, x2_train_df, gx1, fname, dir_name='output_'+self.modelname)
save_info.save_scores(epoch, x1_train_df, x2_train_df, gx1, training_metrics, fname, dir_name='output_'+self.modelname)
#save_plots.plot_progress(epoch, x1_train_df, x2_train_df, gx1, training_metrics, fname, umap=False,
# dir_name='figures_'+self.modelname, autoencoder=False, modelname=self.modelname)
if __name__ == '__main__':
# sample5, sample75, sample65
import os
from loading_and_preprocessing.data_loader import load_data_basic
import argparse
parser = argparse.ArgumentParser()
parser.add_argument('sample', type=str)
parser.add_argument('path', type=str)
args = parser.parse_args()
sample_name = 'sample' + args.sample
#path = '/cluster/home/hthrainsson/chevrier_data_pooled_full_panels.parquet'
#path = r'C:\Users\heida\Documents\ETH\Deep Learning\chevrier_data_pooled_full_panels.parquet'
x1_train, x1_test, x2_train, x2_test = load_data_basic(args.path, sample=sample_name,
batch_names=['batch1', 'batch3'], seed=42, panel=None,
upsample=True)
gan = GAN('residual_gan_full_panels', x1_train.shape[1])
gan.train(x1_train, x2_train, epochs=600, batch_size=64, sample_interval=25)
training_metrics["epoch"].append(epoch)
training_metrics["d_loss"].append(d_loss[0])
training_metrics["d_accuracy"].append(d_loss[1])
training_metrics["g_loss"].append(g_loss)
# If at save interval => save generated image samples
if epoch % sample_interval == 0:
print('generating plots and saving outputs')
gx1 = self.generator.predict(x1_train_df)
self.generator.save(os.path.join('models_' + self.modelname, fname, 'generator' + str(epoch)))
save_info.save_dataframes(epoch, x1_train_df, x2_train_df, gx1, fname, dir_name='output_'+self.modelname)
save_info.save_scores(epoch, x1_train_df, x2_train_df, gx1, training_metrics, fname, dir_name='output_'+self.modelname)
#save_plots.plot_progress(epoch, x1_train_df, x2_train_df, gx1, training_metrics, fname, umap=False,
# dir_name='figures_'+self.modelname, autoencoder=False, modelname=self.modelname)
if __name__ == '__main__':
import os
from loading_and_preprocessing.data_loader import load_data_basic
path = '/cluster/home/hthrainsson/chevrier_data_pooled_full_panels.parquet'
# path = r'C:\Users\heida\Documents\ETH\Deep Learning\chevrier_data_pooled_full_panels.parquet'
x1_train, x1_test, x2_train, x2_test = load_data_basic(path, sample='sample5', batch_names=['batch1', 'batch3'],
seed=42, panel=None, upsample=True)
gan = GAN('residual_gan_wo_residuals_full_panels', x1_train.shape[1])
gan.train(x1_train, x2_train, epochs=600, batch_size=64, sample_interval=25)
x1_train, x1_test, x2_train, x2_test = load_data_basic(path, sample='sample10', batch_names=['batch1', 'batch3'],
seed=42, panel=None, upsample=True)
gan = GAN('residual_gan_wo_residuals_full_panels', x1_train.shape[1])
gan.train(x1_train, x2_train, epochs=600, batch_size=64, sample_interval=25)
gx1,
plot_model,
fname,
umap=False)
return plot_model
if __name__ == '__main__':
import os
from loading_and_preprocessing.data_loader import load_data_basic, load_data_cytof
# path = r'C:\Users\heida\Documents\ETH\Deep Learning\2019_DL_Class_old\code_ADAE_\chevrier_data_pooled_panels.parquet'
path = r'C:\Users\Public\PycharmProjects\deep\Legacy_2019_DL_Class\data\chevrier_data_pooled_panels.parquet'
# x1_train, x1_test, x2_train, x2_test = load_data_cytof(path, patient_id='rcc7', n=10000)
x1_train, x1_test, x2_train, x2_test = load_data_basic(
path,
sample='sample5',
batch_names=['batch1', 'batch2'],
seed=42,
panel='tcell')
# path = r'C:\Users\Public\PycharmProjects\deep\2019_DL_Class\loading_and_preprocessing'
# path = path + '/toy_data_gamma_small.parquet' # '/toy_data_gamma_large.parquet'
#x1_train, x1_test, x2_train, x2_test = load_data_basic(path, patient='sample1', batch_names=['batch1', 'batch2'],
# seed=42, n_cells_to_select=0)
gan = GAN(x1_train.shape[1])
gan.train(x1_train,
x2_train,
epochs=3000,
batch_size=64,
sample_interval=50)
gx1,
training_metrics,
fname,
dir_name='output_' + self.modelname,
model_description=model_description)
# save_plots.plot_progress(epoch, x1_train_df, x2_train_df, gx1, plot_model, fname,
# dir_name='output_' + self.modelname)
return plot_model
if __name__ == '__main__':
import os
from loading_and_preprocessing.data_loader import load_data_basic, load_data_cytof
path = '..\data\chevrier_samples_5_65_75.parquet'
samples = ['sample5', 'sample65', 'sample75']
modelname = 'gan_autoencoder_diamond_narrower_full'
for s in samples:
x1_train, x1_test, x2_train, x2_test = load_data_basic(
path,
sample=s,
batch_names=['batch1', 'batch3'],
seed=42,
panel=None)
gan = GAN(x1_train.shape[1], modelname)
gan.train(x1_train,
x2_train,
epochs=3000,
batch_size=64,
sample_interval=50)
#!/usr/bin/env python
import sys
import os
sys.path.append(os.path.dirname(sys.path[0]))
from vanilla_gan.gan_batches_optimized import GAN
from loading_and_preprocessing.data_loader import load_data_basic
path = 'data\chevrier_samples_5_65_75.parquet'
sample_names = ['sample5', 'sample65', 'sample75']
batch_names = ['batch1', 'batch3']
for sample_name in sample_names:
x1_train, x1_test, x2_train, x2_test = load_data_basic(path, sample=sample_name,
batch_names=batch_names, seed=42, panel=None,
upsample=True)
gan = GAN(x1_train.shape[1], modelname='gan_vanilla_full')
gan.train(x1_train, x2_train, epochs=1000, batch_size=64, sample_interval=50)
