def test_blacklist_index_fname_to_blacklist_bool_list(self):
ind_bl_fname = copy(self.cfg['blacklist_fname']).replace("bool", "ind")
out_bl_fname = self.cfg['blacklist_fname'].replace("bool", "bool_test")
assert "all_fnames" in self.cfg
assert self.cfg["all_fnames"]
dataio.blacklist_index_fname_to_blacklist_bool_list(
ind_bl_fname, out_bl_fname, self.cfg["all_fnames"][0])
bl = dataio.get_ok_nodes(out_bl_fname)
assert type(bl) is np.ndarray
assert bl.dtype == bool
# specific for the blacklist provided in the testdata:
assert not bl[0]
assert not bl[-1]
os.remove(out_bl_fname)
assert not os.path.exists(out_bl_fname)
def test_modules_igraph(self):
tag = "multilevel"
compcoms.comp_communities_igraph(self.cfg, tag)
# viz individual
individual = 0
viz.viz_com_stru_for_ind(
self.cfg['all_fnames'][individual],
self.cfg
)
consensus_out_fname = (self.cfg['outdata_dir'] +
tag +
"_" + "all_fnames" + "_" +
str(self.cfg['density']) +
".pkl")
compcoms.comp_consensus_partition(
self.cfg,
'all_fnames',
consensus_out_fname,
comdet_tag=tag
)
data = dataio.load(consensus_out_fname)
partition = data[tag]
filtered_partition = partition[
dataio.get_ok_nodes(self.cfg['blacklist_fname'])
]
fig = viz.viz_com_structure_using_slices(filtered_partition, self.cfg)
fig.savefig(consensus_out_fname.replace(".pkl", ".pdf"), format="pdf")
fig = viz.comp_and_viz_cluster_diff_matrices(
partition,
self.cfg,
self.cfg['all_fnames'],
len(self.cfg['group_1_mat_fnames']),
vmin=-5,
vmax=5,
suptitle="",
recompute=True
)
fig.savefig(consensus_out_fname.replace(".pkl", "_coarse-grained.pdf"))
comdet_tag=comdet_tag)
print "Matching modules algorithmically"
consensus_clu_movie = \
dataio.load_pickle(consensus_fname0)[comdet_tag]
consensus_clu_rest = \
dataio.load_pickle(consensus_fname1)[comdet_tag]
consensus_clu_movie, consensus_clu_rest, n_clu_1, n_clu_2 = \
viz._get_matched_renumbered_communities_for_viz(
consensus_clu_movie, consensus_clu_rest
)
uf_consensus_clus = [consensus_clu_movie, consensus_clu_rest]
print "Setting colors for modules"
ok_nodes = dataio.get_ok_nodes(cfg['blacklist_fname'])
filtered_consensus_clus = [
consensus_clu_movie[ok_nodes], consensus_clu_rest[ok_nodes]
]
n_cols = np.maximum(np.max(consensus_clu_movie),
np.max(consensus_clu_rest)) + 1
# +1 because indexing starts from zero
colors_list = np.array([
"#e41a1c", # red
"#f781bf", # pink
"#377eb8", # blue
"#95c3dd", # light blue
"#1f8f61", # dark green
"#a6dc6c", # light green
def comp_node_pvals(cfg, matlabQValComps=False):
"""
Computes the permutation-based p-values for the precomputed nodeproperties
The comparison is made between
``cfg["group_1_mat_fnames"]`` and ``cfg["group_1_mat_fnames"]``
Parameters
----------
cfg : dict
brainnets config dict
"""
if not cfg['paired']:
assert cfg['n_it_permutation'] != "all", \
"all is not allowed for non-paired tests"
fnames_group1 = cfg['group_1_mat_fnames']
fnames_group2 = cfg['group_2_mat_fnames']
fnames = fnames_group1 + fnames_group2
node_props_data = dataio.merge_and_load_props_data(fnames, "node_props",
cfg['node_props'], cfg)
ok_nodes = dataio.get_ok_nodes(cfg['blacklist_fname'])
for key in node_props_data:
if key != settings.densities_tag:
node_props_data[key] = node_props_data[key][:, ok_nodes]
n1 = len(fnames_group1)
n2 = len(fnames_group2)
node_pval_stats = _comp_perm_test_stats_for_some_props(
node_props_data,
n1,
n2,
cfg['paired'],
cfg['n_it_permutation'],
measures.mean_difference,
matlabQValComps=False)
# print node_pval_stats.keys()
# for nodeProp in settings.node_props:
# if nodeProp == settings.densities_tag:
# continue
# propPValStats = node_pval_stats[nodeProp]
# data = node_pval_stats[nodeProp]
# out_dict[settings.pval_tag] = \
# dataio.expand_1D_node_vals_to_non_blacklisted_array(
# propPValStats[settings.pval_tag], ok_nodes)
# out_dict[settings.meandifference_tag] = \
# dataio.expand_1D_node_vals_to_non_blacklisted_array(
# propPValStats[settings.meandifference_tag], ok_nodes)
# if matlabQValComps:
# out_dict[settings.pfdrs_tag] = \
# dataio.expand_1D_node_vals_to_non_blacklisted_array(
# propPValStats[settings.pfdrs_tag], ok_nodes)
# out_dict[settings.qval_tag] = \
# dataio.expand_1D_node_vals_to_non_blacklisted_array(
# propPValStats[settings.qval_tag], ok_nodes)
# except:
# raise
out_fname = fnc.get_stats_fname(cfg, "node_props")
dataio.save_pickle(out_fname, node_pval_stats)