def comp_link_sim_matrices(cfg):
"""
Compute link similarity matrices for given network densities.
specified in ``cfg['density_range']`` across
``cfg['all_fnames']``
Results are saved in ``cfg['outdata_dir'])``
Parameters
----------
cfg : dict
brainnets config dict
"""
config.require(cfg, [
"all_fnames", "blacklist_fname", "density_range", "include_mst",
"n_cpus"
])
densities = cfg['density_range']
arg_list = zip(densities, [cfg] * len(densities))
link_sim_mat_list = ch.run_in_parallel(_comp_link_sim_mat_worker,
arg_list,
cfg["n_cpus"],
chunksize=1)
out_dict = {
settings.densities_tag: densities,
settings.common_links_tag: link_sim_mat_list,
settings.config_tag: cfg
}
dataio.save_pickle(fnc.get_fname(cfg, settings.common_links_tag), out_dict)
def comp_paired_common_and_differing_link_distances(cfg):
"""
For each pair, computes the distances of common and differing
links.
"""
config.require(cfg, [
"group_1_mat_fnames", "group_2_mat_fnames", "blacklist_fname",
"node_info_fname", "density_range", "n_cpus"
])
fnames_group_1 = cfg['group_1_mat_fnames']
fnames_group_2 = cfg['group_2_mat_fnames']
assert len(fnames_group_1) == len(fnames_group_2)
n = len(fnames_group_1)
arg_list = zip(fnames_group_1, fnames_group_2, [cfg] * n)
ch.run_in_parallel(_paired_common_and_diff_link_distances_worker, arg_list,
cfg['n_cpus'])
def comp_louvain_communities(cfg):
"""
Computes louvain communities for a certain network density for all
correlation matrices in ``cfg['all_fnames']``
The results are saved to the output folder (``cfg['outdata_dir']``)
Currently only the unweighted louvain method is used.
Parameters
----------
cfg : a brainnets config dict
Returns
-------
coms :
the communities as a membershiplist
mods : numpy array
the corresponding values of modularity
"""
config.require(cfg, ['all_fnames', 'blacklist_fname', 'density',
'n_it_comdet', 'include_mst', 'n_cpus'])
argList = [(fname, cfg) for fname in cfg['all_fnames']]
comsAndModularities = ch.run_in_parallel(
_compute_louvain_coms_worker, argList, cfg['n_cpus'])
coms = [comsAndModularities[i][0] for i in range(len(comsAndModularities))]
mods = [comsAndModularities[i][1] for i in range(len(comsAndModularities))]
return coms, mods
def comp_communities_igraph(cfg, com_det_method, com_det_options_dict=None):
"""
Computes communities for a certain network density for all
correlation matrices in ``cfg['all_fnames']``
The results are saved to the output folder (``cfg['outdata_dir']``)
Parameters
----------
cfg : a brainnets config dict
com_det_method: str, or igraph function returning
com_det_options_dict: dict
Returns
-------
coms :
the communities as a membership list
"""
if isinstance(com_det_method, str):
com_det_method = tag_to_igraph_comdet_method(com_det_method)
if com_det_options_dict is None:
com_det_options_dict = {}
config.require(cfg, ['all_fnames',
'blacklist_fname',
'density',
'include_mst',
'n_cpus',
'n_it_comdet'
]
)
arg_list = [(fname, cfg, com_det_method, com_det_options_dict)
for fname in cfg['all_fnames']]
coms = ch.run_in_parallel(_compute_coms_worker, arg_list, cfg['n_cpus'])
return coms
def comp_link_distances(cfg):
"""
Computes all link distances (in MNI space) over a range of network
densities.
Parameters
----------
cfg : dict
the brainnets config dictionary
"""
config.require(cfg, [
"all_fnames", "blacklist_fname", "density_range", "node_info_fname",
"include_mst", "n_cpus"
])
all_fnames = cfg["all_fnames"]
arg_list = zip(all_fnames, [cfg] * len(all_fnames))
ch.run_in_parallel(_link_dist_worker, arg_list, cfg['n_cpus'])
def comp_node_props(cfg):
"""
Computes node properties for ``cfg["all_fnames"]`` for a given network
density
Parameters
----------
cfg : dict
brainnets config dictionary
"""
config.require(cfg, [
"all_fnames", "blacklist_fname", "density", "include_mst", "n_cpus",
"node_props"
])
all_fnames = cfg["all_fnames"]
n = len(all_fnames)
arg_list = zip(all_fnames, [cfg] * n)
ch.run_in_parallel(_node_prop_worker, arg_list, cfg["n_cpus"])
def compute_global_properties(cfg, weighted=False):
"""
Computes global network properties for ``cfg["all_fnames"]`` for a given
network density
Parameters
----------
cfg : dict
brainnets config dictionary
weighted : bool
if True, the network is considered as weighted.
if False, network is unweighted
"""
config.require(cfg, [
"all_fnames", "blacklist_fname", "density_range", "include_mst",
"n_cpus", "global_w_props"
])
n = len(cfg['all_fnames'])
arg_list = zip(cfg['all_fnames'], [cfg] * n)
if weighted:
worker = _global_w_prop_worker
else:
worker = _global_uw_prop_worker
ch.run_in_parallel(worker, arg_list, cfg['n_cpus'])