add_cor_scores

[TingTingShao]

Hi,

I ran into error when running

net = snap.tl.init_network_from_annotation(regions, 
    anno_file =snap.genome.hg38,  
    upstream = 250000,  
    downstream= 250000,  
    id_type = 'gene_name',  
    coding_gene_only = True)
net = snap.tl.add_cor_scores(net, gene_mat=gene_mat, peak_mat=peak_mat, select=None, overwrite=False)

error shows:

net = snap.tl.add_cor_scores(net, gene_mat=gene_mat, peak_mat=peak_mat, select=None, overwrite=False)
if sp.issparse(y):
    y = y.todense()
scores = np.ravel(spearman(X.T, y.reshape((1, -1))))
for nd, sc in zip(nd_X, scores):
    setattr(network.get_edge_data(nd, nd_y), key, sc)

Looking forward to your reply!

Thanks
tingting

[kaizhang]

what was the error message?

[TingTingShao]

Sorry, miss the error line, this is the error:

PanicException: Cannot compute correlation for type uint32

[thecatsmilk]

I am having this same issue. Is there any update on this?

[Habush]

I had the same issue. The source of the issue is the spearman function expects both it's arguments to be floats but the peak_mat data is type uint32. A simple fix is to convert the peak_mat data to float.

from scipy.sparse import csr_matrix, csc_matrix

if isinstance(peak_mat.X, csr_matrix):
    peak_mat.X = csr_matrix(peak_mat.X, dtype=np.float32)
elif isinstance(peak_mat.X< csc_matrix):
    peak_mat.X = csc_matrix(peak_mat.X, dtype=np.float32)