mouse_row = selected_rows.iloc[0]
mouse_row_new = main_source_extraction(mouse_row, parameters_source_extraction, dview)
mouse_row_new = db.set_version_analysis('source_extraction', mouse_row_new)
states_df = db.append_to_or_merge_with_states_df(states_df, mouse_row_new)
db.save_analysis_states_database(states_df, path=analysis_states_database_path, backup_path = backup_path)
#%% COMPONENT EVALUATION
min_SNR = 3 # adaptive way to set threshold on the transient size
r_values_min = 0.85 # threshold on space consistency (if you lower more components
# will be accepted, potentially with worst quality)
parameters_component_evaluation = {'min_SNR': min_SNR,
'rval_thr': r_values_min,
'use_cnn': False}
cm.stop_server(dview=dview)
mouse_row_new = main_component_evaluation(mouse_row_new, parameters_component_evaluation)
component_evaluation_output = eval(mouse_row_new['component_evaluation_output'])
input_hdf5_file_path = component_evaluation_output['main']
cnm = load_CNMF(input_hdf5_file_path)
print('Accepted components = ')
print(len(cnm.estimates.idx_components))
print('Rejected components = ')
print(len(cnm.estimates.idx_components_bad))
selection = selected_rows.query('(session ==' + f'{session}' + ')')
for cropping_version in [1, 2, 3, 4]:
print(cropping_version)
selected_rows = db.select(
states_df,
'component_evaluation',
mouse=mouse_number,
session=session,
is_rest=is_rest,
decoding_v=decoding_version,
cropping_v=cropping_version,
motion_correction_v=motion_correction_version,
alignment_v=alignment_version,
source_extraction_v=source_extraction_version)
mouse_row = selected_rows.iloc[0]
mouse_row_new = main_component_evaluation(
mouse_row, parameters_component_evaluation, session_wise=True)
states_df = db.append_to_or_merge_with_states_df(
states_df, mouse_row_new)
db.save_analysis_states_database(states_df,
path=analysis_states_database_path,
backup_path=backup_path)
##########
decoding_version = 1
motion_correction_version = 1
for session in [1, 2, 4]:
print(session)
# Run decoding for group of data tha have the same cropping parameters (same mouse)
selection = selected_rows.query('(session ==' + f'{session}' + ')')
#%%
##To see components one by one in browser
#cnm.estimates.nb_view_components(img=cn_filter, idx=cnm.estimates.idx_components, cmap = 'gray')
#select the number of components to plot
idx_array = np.arange(10)
#plot components (non deconvolved)
fig = src.steps.source_extraction.get_fig_C_stacked(cnm.estimates.C,
idx_components=idx_array)
#%%
min_SNR = 10 # adaptive way to set threshold on the transient size
r_values_min = 0.99 # threshold on space consistency (if you lower more components
# will be accepted, potentially with worst quality)
parameters_component_evaluation = {
'min_SNR': min_SNR,
'rval_thr': r_values_min,
'use_cnn': False
}
main_component_evaluation(index, row, parameters_component_evaluation)
#%%
#load cnmf class with evaluation
component_evaluation_output = eval(row.loc['component_evaluation_output'])
input_hdf5_file_path = component_evaluation_output['main']
cnm = load_CNMF(input_hdf5_file_path)
print(len(cnm.estimates.idx_components))