Cytoscape network view problem: running out of memory

Question

[piyushjo]

Hi,

I am running ARACNe plugin in cytoscape to find correlated genes using mutual inference algorithm. I upload the RNA-Seq data and run the complete process. The information bar says calculating MI and then gets stuck at network view. I have a workstation with 47 GB ram. The MI calculation part takes upto 27 GB, then when the information box says building network view, the memory consumption slowly increase to 47 GB and still i don't see a network graph. Is there a way I can fix it? Or at least get the MI matrix.

Thanks!

[Kevin Blighe]

What are the dimensions of your input data matrix?

[piyushjo]

The .csv file has 14,999 genes with two repeats. Is there a way I can generate the MI matrix somewhere and then visualize on cytoscape? I am interested in connecting TF with lncrna so I can't use the network from string.

[Kevin Blighe]

That's a lot of genes. Could you, perhaps, modify the thresholds -t and -p ?

[piyushjo]

HI Kevin,

Thanks for your suggestion. I tried lowering down the number of genes. I realized the problem I was having was number of genes and also because I was using two RNA-Seq repeats of sample cell line, so everything was correlated. I tried minet package from R on some patient samples and was able to plot the network using cytoscape. Do you have suggestion regarding any R package that you have used for MI, that you trust the analysis. I want to compare the my analysis to make sure I am following proper analysis.

Thanks!