Avoiding the empty gray box

I am a new user of HiGlass, and I am trying to use it to visualize an arbitrary numpy matrix, following the example given in the “Jupyter examples” page. When I first loaded my matrix, I just got an empty viewport, like this:

After experimenting for a bit, I learned that it was possible for me see the matrix if I kept only a small submatrix and rescaled the values to a much smaller range. Here is what I get:

I achieved this result by extracting a submatrix of size 2000x2000 and raising the entire matrix to the power 0.1. I find that if I use all of the 82548 rows of the matrix or if I expand the range of values to the original range (0 to 2166895104), then I am back to the grey screen. Selecting “zoom to extent” doesn’t solve the problem.

Is there something I can do to be able to visualize this entire matrix?

Here’s the code:

import scipy.sparse as sparse 

compressed_matrix_file="DIA-S4-1.npz"
noisy_sparse = sparse.load_npz(compressed_matrix_file).tocsr()
# load the whole matrix
noisy = noisy_sparse.toarray()
# load part of the matrix
rt_min = 0
rt_max = 2000
mz_min = 38000
mz_max = 40000
noisy = noisy_sparse.toarray()[rt_min:rt_max]
noisy = np.transpose(noisy)[mz_min:mz_max]
print(noisy.shape)
noisy = np.power(noisy, 1/10)
print(noisy.min(), noisy.max())

ts = Tileset(
    tileset_info=lambda: npmatrix.tileset_info(noisy),
    tiles=lambda tids: npmatrix.tiles_wrapper(noisy, tids)
)

display, server, viewconf = higlass.display([
    View([
        Track(track_type='top-axis', position='top'),
        Track(track_type='left-axis', position='left'),
        Track(track_type='heatmap',
              position='center',
              tileset=ts,
              height=250),

    ])
])
display

DIA-S4-1.npz.gz

[wsnoble]

Yes, after I installed the new version it works. Thanks!

[pkerpedjiev]

Interesting… you must have clodius installed because it’s a dependency of higlass-python. And the tiles function here, npmatrix.tileset_info comes from the clodius package.

    tileset_info=lambda: npmatrix.tileset_info(noisy),

Are you using the same virtual environment for your notebook and the place you tried installing clodius?

I just issued a new release so you can now do pip install clodius>=0.17.1 to get the latest version.