

In [ ]:

    
# Author: Praveen Sripad <pravsripad@gmail.com>

import numpy as np

import mne
from mne.datasets import sample
from jumeg.connectivity.con_utils import make_communities

from nilearn import plotting
import nibabel as nib

import matplotlib.pyplot as plt

data_path = sample.data_path()
subjects_dir = data_path + '/subjects'
subject = 'fsaverage'
aparc = mne.read_labels_from_annot(subject, subjects_dir=subjects_dir,
                                   parc='aparc')

# make a random connectivity matrix
np.random.seed(42)
con = np.random.random((34, 34))
con[con < 0.8] = 0.

top_nodes_list, n_communities = make_communities(con)

# FIXME better colouring
# cmaps_list = ['Pastel1', 'Pastel2', 'Paired', 'Accent',
#               'Dark2', 'Set1', 'Set2', 'Set3',
#               'tab10', 'tab20', 'tab20b', 'tab20c']
# color_list = plt.cm.Pastel1(np.linspace(0, 1, len(top_nodes_list)))

color_list = ['red', 'orange', 'green', 'purple', 'navy', 'blue']
cmaps_list = ['Reds', 'Oranges', 'Greens', 'Purples', 'PuBu', 'Blues']

fig = plt.figure(facecolor='w', edgecolor='w')
# fig = None

for top_nodes, col, cmap in zip(top_nodes_list, color_list, cmaps_list):

    coords = []
    community_ = [aparc[tp] for tp in top_nodes]

    for lab in community_:
        if lab.name is 'unknown-lh':
            continue
        print(lab.name)
        # get the center of mass
        com = lab.center_of_mass('fsaverage')
        if lab.name.endswith('lh'):
            # obtain mni coordinated to the vertex from left hemi
            coords_ = mne.vertex_to_mni(com, hemis=0, subject=subject,
                                        subjects_dir=subjects_dir)[0]
        else:
            coords_ = mne.vertex_to_mni(com, hemis=1, subject=subject,
                                        subjects_dir=subjects_dir)[0]
        coords.append(coords_)

    n_nodes = np.array(coords).shape[0]
    # make a dummy connectivity matrix
    # FIXME use the existing con matrix with indices chosen instead
    con_ = np.ones((n_nodes, n_nodes)) / 2.
    con_[np.diag_indices(n_nodes)] = 0.
    con_[np.triu_indices(n_nodes, k=1)] = 0.
    con_ += con_.T

    # plot the connectome on a glass brain background
    plotting.plot_connectome(con_, coords, node_color=col, edge_cmap=cmap,
                             figure=fig, edge_vmin=0., edge_vmax=1.)

plt.show()
# plt.savefig('communities.png')
Plot Networkx Communities on a connectome plot.
