search

brianhie / trajectorama

Multi-study integration of cellular trajectories
http://trajectorama.csail.mit.edu
MIT License
18 stars 3 forks source link

Thread getting Killed #7

Open TheInfamousWayne opened 4 years ago

TheInfamousWayne commented 4 years ago

Terminal Output:-

. . . computing neighbors computing neighbors computing neighbors finished: added to .uns['neighbors'] 'distances', distances for each pair of neighbors 'connectivities', weighted adjacency matrix (0:01:47) running Louvain clustering using the "louvain" package of Traag (2017) finished (0:00:01) finished: added to .uns['neighbors'] 'distances', distances for each pair of neighbors 'connectivities', weighted adjacency matrix (0:01:49) running Louvain clustering using the "louvain" package of Traag (2017) finished: added to .uns['neighbors'] 'distances', distances for each pair of neighbors 'connectivities', weighted adjacency matrix (0:01:49) running Louvain clustering finished (0:00:00) using the "louvain" package of Traag (2017) finished (0:00:00) Killed

TheInfamousWayne commented 4 years ago

My code is running for some time. Eventually, I'm just getting a "thread killed" error message. I'm running the following code: `` import scanpy as sc import pandas as pd import numpy as np import trajectorama from anndata import AnnData import matplotlib.pyplot as plt

plt.switch_backend('agg') sc.settings.set_figure_params(scanpy=True, dpi=100, dpi_save=200, format='pdf') sainath=sc.settings.figdir sc.settings.verbosity = 3 sc.logging.print_versions() result_file='./write/sainath.h5ad'

sc.settings.set_figure_params(dpi=100) bdata = sc.read_h5ad("/home/vaibhav_agrawal51_va/mo_new_data_after_RV_6_improved.h5ad") X = bdata.X studies = list(bdata.obs.condition) Xs_coexpr, sample_idxs = trajectorama.transform( X, studies, corr_cutoff=0.7, corr_method='spearman', cluster_method='louvain', min_cluster_samples=500, )

n_features = X.shape[1] triu_idx = np.triu_indices(n_features) # Indices of upper triangle. X_coexpr = np.concatenate([ Xs_coexpr[triu_idx].flatten() for Xs_coexpr in Xs_coexpr ])

Plot KNN graph in coexpression space.

adata = AnnData(X_coexpr) sc.pp.neighbors(adata) sc.tl.draw_graph(adata) sc.pl.draw_graph(adata,save="trajectory") ``

brianhie commented 4 years ago

Perhaps it's killed due to memory usage? Can you try using sparse matrices, e.g.,

import scipy.sparse
X_coexpr = scipy.sparse.csr_matrix(X_coexpr)