Marker size array is not well displayed in plot_glass_brain

[robbisg]

What version of Nilearn are you using?

In [36]: nilearn.__version__ 
Out[36]: '0.6.0a'

What were you trying to do?
Plot a connectome using plot_connectome function, giving the a numpy array as node_size.

Here's the snippet to replicate:
coords and colors are obtained by some functions I've implemented in my libraries.

from nilearn.plotting import plot_connectome
matrix = np.random.rand(90,90)
matrix = (matrix * matrix.T)/2
t_matrix = matrix * np.int_(matrix > 0.45) #Plot only some connections
f = plot_connectome(t_matrix, 
                        coords[:90], 
                        colors[:90], 
                        t_matrix.sum(0)*100, 
                        'magma_r',
                        display_mode='lyr',
                        edge_vmin=0.45,
                        edge_vmax=0.5,
                        figure=pl.figure(figsize=(25,15)),
                        )

What did you expect will happen?

What actually happened?
The node size of left and right images are different from those obtained in coronal view.

I've found the problem in displays.pyand fixed it by adding at line 417
Maybe we also need some control on array/list length.

            if isinstance(marker_size, np.ndarray) or \
                isinstance(marker_size, list):
                marker_size = np.array(marker_size)[relevant_coords]

I can do the PR, if you are ok.

[kchawla-pi]

Yes please!

…

On Thu, Oct 17, 2019, 18:18 robbisg ***@***.***> wrote: What version of Nilearn are you using? In [36]: nilearn.__version__ Out[36]: '0.6.0a' What were you trying to do? Plot a connectome using plot_connectome function, giving the a numpy array as node_size. Here's the snippet to replicate: coords and colors are obtained by some functions I've implemented in my libraries. from nilearn.plotting import plot_connectome matrix = np.random.rand(90,90) matrix = (matrix * matrix.T)/2 t_matrix = matrix * np.int_(matrix > 0.45) #Plot only some connections f = plot_connectome(t_matrix, coords[:90], colors[:90], t_matrix.sum(0)*100, 'magma_r', display_mode='lyr', edge_vmin=0.45, edge_vmax=0.5, figure=pl.figure(figsize=(25,15)), ) What did you expect will happen? [image: fixed] <https://user-images.githubusercontent.com/695037/67026948-0d33a980-f109-11e9-83f5-24c2bf1c66e1.png> What actually happened? The node size of left and right images are different from those obtained in coronal view. [image: buggy] <https://user-images.githubusercontent.com/695037/67026961-13298a80-f109-11e9-909b-f8565edcd422.png> I've found the problem in displays.pyand fixed it by adding at line 417 Maybe we also need some control on array/list length. if isinstance(marker_size, np.ndarray) or \ isinstance(marker_size, list): marker_size = np.array(marker_size)[relevant_coords] I can do the PR, if you are ok. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#2185?email_source=notifications&email_token=AB6SXRAZVHXRL5LFDHWOGT3QPCF5FA5CNFSM4JB37HLKYY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4HSQL4QA>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AB6SXRBA5TAH3YLIWECTRNLQPCF5FANCNFSM4JB37HLA> .

[jeromedockes]

I think you are right and a PR would be welcome. Thanks!