[MRG] Allow asymmetric matrices in plot_connectome #2703
Conversation
Codecov Report
@@ Coverage Diff @@
## master #2703 +/- ##
==========================================
- Coverage 87.59% 87.58% -0.01%
==========================================
Files 100 100
Lines 13171 13155 -16
Branches 2555 2559 +4
==========================================
- Hits 11537 11522 -15
+ Misses 1026 1025 -1
Partials 608 608
Continue to review full report at Codecov.
|
|
LGTM, thx ! |
|
Azure failure related to empty arrows crashing for matplotlib pre 3.1 (see: matplotlib/matplotlib#13658) |
86ab429 to
e1ed9ec
Compare
This might be too simplistic, but maybe something along those lines ? import numpy as np
from nilearn import datasets, plotting
from nilearn.input_data import NiftiLabelsMasker
def lag_correlation(time_series, lag):
"""Basic home-made lag correlation function."""
n_subjects = len(time_series)
n_samples, n_features = time_series[0].shape
lag_cor = np.zeros((n_subjects, n_features, n_features))
for subject,serie in enumerate(time_series):
for i in range(n_features):
for j in range(n_features):
lag_cor[subject, i, j] = np.corrcoef(serie[lag:, i],
serie[:-lag, j])[0, 1]
return np.mean(lag_cor, axis=0)
# Get atlas and func data
yeo = datasets.fetch_atlas_yeo_2011()
data = datasets.fetch_development_fmri(n_subjects=10)
# Transform into time series
masker = NiftiLabelsMasker(labels_img=yeo['thick_17'], standardize=True,
memory='nilearn_cache')
time_series = []
for func, confounds in zip(data.func, data.confounds):
time_series.append(masker.fit_transform(func, confounds=confounds))
# Compute node coordinates
coordinates = plotting.find_parcellation_cut_coords(labels_img=yeo['thick_17'])
# Compute lag correlation matrices and plot connectomes
for lag in [1, 10]:
lag_cor = lag_correlation(time_series, lag)
plotting.plot_connectome(lag_cor, coordinates, edge_threshold="90%",
title='Lag{} correlation'.format(lag))
|
|
Thx for the example suggestion. |
Thanks for the comment! I agree and I updated one of the connectome-related examples accordingly. |
|
At least it looks OK to me. |
| time_series) | ||
| mean_empirical_correlation_matrix = empirical_connectome_measure.mean_ | ||
| lag0_correlation = lag_correlation(time_series, 0) | ||
| assert_array_almost_equal(mean_empirical_correlation_matrix, |
There was a problem hiding this comment.
This is a bit weird in an example.
There was a problem hiding this comment.
I agree, this is probably not super useful in an example.
I simplified this section (see d2e3b1c)
| for direction in ['z', 'r', 'y']: | ||
| assert(len([patch for patch in ax.axes[direction].ax.patches | ||
| if isinstance(patch, FancyArrow)]) == | ||
| np.prod(adjacency_matrix.shape) - 2) |
There was a problem hiding this comment.
This is not related to this PR ? I'm confused.
There was a problem hiding this comment.
I am testing that we get the correct number of arrows in the connectome plot when setting some (2) elements of the matrix to zero and for a different display mode. If this is too redundant with other tests added in this PR, I can delete it
4d67809 to
d2e3b1c
Compare
|
Thanks for the reviews! |
* 'master' of github.com:nilearn/nilearn: Release 0.7.1 (nilearn#2727) Add two_sided flag to get_clusters_table (nilearn#2719) bugfix resample only if type is not contour or contourf (nilearn#2656) Support atlases in plot_carpet (nilearn#2702) Fix whatsnew entry with public function reference (nilearn#2716) EHN signal.clean standardise option PSC to allow negative mean signal (nilearn#2714) Replace NaNs with zeros in NiftiLabelsMasker (nilearn#2712) [MRG] Allow asymmetric matrices in plot_connectome (nilearn#2703) [MRG] Mosaic mode of displaying brain images (nilearn#2684) Resampling functionality for NeuroVault (nilearn#2696) Solving "NiftiMapsMasker and NiftiLabelsMasker cannot transform 3D images" (nilearn#2708) Raise warning for FirstLevelModel fit when confounds and design_matrices are both specified (nilearn#2710) Fixed numpy warning when img has NaNs. (nilearn#2709) [MRG] High variance confounds in maskers (nilearn#2697) fix sampling (nilearn#2706) [MRG] Handle nans in surf plotting (nilearn#2691) [FIX] get_clusters_table: ensure input is 3D and accept filenames (nilearn#2701) Warn in nifti masker report when no image provided to fit (nilearn#2694) fix for FirstLevelModel compute_contrast for multiple runs with varying matrice size or columns (nilearn#2688)
Addresses #1513
Description
This PR proposes to allow asymmetric matrices to be plotted as a directed graph through the
plot_connectomefunction. Previous behaviour was to raise an error.Example
Todo