def apply_parcellation(subject):
'''
Applies parcellation for all sparsity levels
Parameters
------
subject - int,
subject id
'''
average_desikan = np.load(
'/data01/ayagoz/sparse_32_concon_HCP/parcellations/desikan_aparc_average_1113.npy',
allow_pickle=True)
source = '/data01/ayagoz/sparse_32_concon_HCP/connectomes/ConCon_resolution'
parcellation = '/data01/ayagoz/sparse_32_concon_HCP/parcellations/connectivity_parcellation_level'
target_folder = '/data01/ayagoz/sparse_32_concon_HCP/connectomes/individual_louvain_parcellation'
for sparsity in range(10, 101, 10):
try:
adj = load_concon(f'{source}/{sparsity}/{subject}.npz',
labels_to_drop=average_desikan)
# print(f'{source}/{sparsity}/{subject}.npz')
for i in [1, 2, 3]:
labels_parcellation = np.load(
f'{parcellation}{i}/{sparsity}/{subject}.npy')
# print(f'{parcellation}{i}/{sparsity}/ensemble_{i}_{sparsity}.npy')
adj_parcellation = squeeze_matrix(adj, labels_parcellation)
np.save(f'{target_folder}/level{i}/{sparsity}/{subject}.npy',
adj_parcellation)
# print(f'{target_folder}/level{i}/{sparsity}/{subject}.npy')
except BaseException as e:
print(e, subject, sparsity, plevel)
def apply_parcellation(subject, parcellation_folder, source_folder,
target_folder):
'''
Applies parcellation for all sparsity levels
Parameters
------
subject - int,
subject id
parcellation_folder - str,
folder with labels to apply
source_folder - str,
folder with concon sparsity folders
target_folder - str,
folder to save to
'''
for sparsity in range(10, 101, 10):
try:
adj = load_concon(f'{source_folder}/{sparsity}/{subject}.npz')
labels_parcellation = np.load(
f'{parcellation_folder}/{subject}.npy')
adj_parcellation = squeeze_matrix(adj,
labels_parcellation,
drop_minus_1=True)
np.save(f'{target_folder}/{sparsity}/{subject}.npy',
adj_parcellation)
except BaseException as e:
print(e, subject)
def get_partitions_at_levels(subject_id, labels_to_drop):
for sparsity in tqdm(range(10, 101, 10)):
try:
path = f'/data01/ayagoz/sparse_32_concon_HCP/ConCon_resolution/{sparsity}/{subject_id}.npz'
adj = load_concon(path, labels_to_drop=labels_to_drop)
partition_level_1 = compute_partition(adj)
partition_level_2 = cluster_subgraphs(adj, partition_level_1, 5,
200)
partition_level_3 = cluster_subgraphs(adj, partition_level_2, 5,
360)
parcellation_path = '/data01/ayagoz/sparse_32_concon_HCP/parcellations'
np.save(
f'{parcellation_path}/connectivity_parcellation_level1/{sparsity}/{subject_id}.npy',
partition_level_1)
np.save(
f'{parcellation_path}/connectivity_parcellation_level2/{sparsity}/{subject_id}.npy',
partition_level_2)
np.save(
f'{parcellation_path}/connectivity_parcellation_level3/{sparsity}/{subject_id}.npy',
partition_level_3)
except BaseException as e:
print(e, subject_id)
def apply_parcellation(subject, parcellation_folder, source_folder,
target_folder):
'''
Applies parcellation for all sparsity levels
Parameters
------
subject - int,
subject id
parcellation_folder - str,
folder with labels to apply
source_folder - str,
folder with concon sparsity folders
target_folder - str,
folder to save to
'''
plevel = parcellation_folder[-1]
average_desikan = np.load(
'/data01/ayagoz/sparse_32_concon_HCP/parcellations/desikan_aparc_average_1113.npy',
allow_pickle=True)
for sparsity in range(10, 101, 10):
try:
adj = load_concon(f'{source_folder}/{sparsity}/{subject}.npz',
labels_to_drop=average_desikan)
labels_parcellation = np.load(
f'{parcellation_folder}/{sparsity}/ensemble_{plevel}_{sparsity}.npy'
)
adj_parcellation = squeeze_matrix(adj, labels_parcellation)
np.save(f'{target_folder}/{sparsity}/{subject}.npy',
adj_parcellation)
except BaseException as e:
print(e, subject, sparsity, plevel)
def apply_parcellation(subject):
'''
Applies parcellation for all sparsity levels
Parameters
------
subject - int,
subject id
'''
concon_folder = '/data01/ayagoz/sparse_32_concon_HCP/connectomes/ConCon_resolution'
aver_path = f'/data01/ayagoz/sparse_32_concon_HCP/parcellations/ensemble_parcellation/average_network_partition/level{plevel}'
cspa_path = f'/data01/ayagoz/sparse_32_concon_HCP/parcellations/ensemble_parcellation/CSPA/level{plevel}'
aver_target = f'/data01/ayagoz/sparse_32_concon_HCP/connectomes/Ensemble_parcellation/Aver_level{plevel}'
cspa_target = f'/data01/ayagoz/sparse_32_concon_HCP/connectomes/Ensemble_parcellation/CSPA_level{plevel}'
average_desikan = np.load(
'/data01/ayagoz/sparse_32_concon_HCP/parcellations/desikan_aparc_average_1113.npy',
allow_pickle=True)
for sparsity in range(10, 101, 10):
try:
adj = load_concon(f'{concon_folder}/{sparsity}/{subject}.npz',
labels_to_drop=average_desikan) #
labels_parcellation = np.load(
f'{aver_path}/aver_level{plevel}_{sparsity}.npy') #
adj_parcellation = squeeze_matrix(adj, labels_parcellation) #
np.save(f'{aver_target}/{sparsity}/{subject}.npy',
adj_parcellation)
labels_parcellation = np.load(
f'{cspa_path}/CSPA_{plevel}_{sparsity}.npy') #
adj_parcellation = squeeze_matrix(adj, labels_parcellation) #
np.save(f'{cspa_target}/{sparsity}/{subject}.npy',
adj_parcellation)
except BaseException as e:
print(e, subject, sparsity, plevel)
if __name__ == "__main__":
random_state = int(argv[1])
for r in tqdm(range(random_state, random_state + 10)):
random = np.random.RandomState(random_state)
sparsity = 10
average_desikan = np.load(
'/data01/ayagoz/sparse_32_concon_HCP/parcellations/desikan_aparc_average_1113.npy',
allow_pickle=True)
paths = glob(
f'/data01/ayagoz/sparse_32_concon_HCP/connectomes/ConCon_resolution/{sparsity}/*.npz'
)
random.shuffle(paths)
adj = load_concon(paths[0], labels_to_drop=average_desikan)
for file in tqdm(paths[1:50]):
adj_temp = load_concon(file, labels_to_drop=average_desikan)
adj += adj_temp
partition_level_1 = compute_partition(adj)
partition_level_2 = cluster_subgraphs(adj, partition_level_1, 5, 200)
partition_level_3 = cluster_subgraphs(adj, partition_level_2, 5, 360)
np.save(
f'/home/kurmukov/subject_stability/aver_50_level1_{sparsity}_{r}.npy',
partition_level_1)
np.save(
f'/home/kurmukov/subject_stability/aver_50_level2_{sparsity}_{r}.npy',
partition_level_2)