Search
Question: WGNA export to external software
0
gravatar for je.maxfield
7 weeks ago by
je.maxfield0 wrote:

I am trying to export my network to cytoscape for visualization. I have a very large dataset. When I run a subset of my data on my personal computer it works just fine, making the files necessary for cytoscape. I can't run the program with my whole dataset on my PC and so I run it on my universities cluster. After about 5hr I get the error: 

*** caught segfault ***

address 0x7f559183f610, cause 'memory not mapped'

Traceback:

1: .Call("tomSimilarity_call", datExpr, as.integer(corTypeC), as.integer(netwo$

2: TOMsimilarityFromExpr(datExpr, power = 3)

An irrecoverable exception occurred. R is aborting now ...

/global/opt/slurm_state/slurmd/job2756987/slurm_script: line 14: 124588 Segment$

 

It it a memory issue with recalculating the topo overlap? Is there a way around this? In the tutorial it says "Recalculate topological overlap if needed", what is the alternative?

the output I get is:

Allowing multi-threading with up to 40 threads.

TOM calculation: adjacency..

..will use 40 parallel threads.

Fraction of slow calculations: 0.000000

..connectivity..

 

The script I am running:

 

library(WGCNA);

options(stringsAsFactors = FALSE);

allowWGCNAThreads(); 

lnames = load(file = "Lda-TRAIT_dataInput.RData");

lnames=load(file = "Lda-02-networkConstruction-block.RData");

TOM = TOMsimilarityFromExpr(datExpr, power = 3);

annot = read.csv(file = "Lda_Genes_DONE.csv");

module =  c("lavenderblush3", "bisque4", "mediumorchid", "darkmagenta", "grey");

probes = names(datExpr)

inModule = is.finite(match(moduleColors, modules));

modProbes = probes[inModule];

modGenes = annot$Gene_ID[match(modProbes, annot$Contig)];

modTOM = TOM[inModule, inModule];

dimnames(modTOM) = list(modProbes, modProbes)

cyt = exportNetworkToCytoscape(modTOM,

   edgeFile = paste("CytoscapeInput-edges-", paste(module, collapse="-"), ".txt$

   nodeFile = paste("CytoscapeInput-nodes-", paste(module, collapse="-"), ".txt$

weighted = TRUE,

   threshold = 0.02,

   nodeNames = modProbes,

   altNodeNames = modGenes,

   nodeAttr = moduleColors[inModule]);

 

Thank you!! 

Jessica

 

ADD COMMENTlink modified 7 weeks ago by Peter Langfelder1.3k • written 7 weeks ago by je.maxfield0
0
gravatar for Peter Langfelder
7 weeks ago by
United States
Peter Langfelder1.3k wrote:

It is most likely a bug in WGCNA compiled code. How big is your data?

If you can, please download and install on your cluster this WGCNA development version in which the bug may be fixed:

https://www.dropbox.com/s/hl49yc0l58nkp36/WGCNA_1.61-84.tar.gz?dl=0

Download the file and in R type install.packages("path/to/WGCNA_1.61-84.tar.gz") - this will install the package in your personal R library on the cluster. Then, when you run library(WGCNA), make sure the greeting message reports version 1.61-84.

My problem in debugging this error is that on our cluster I am unable to reproduce these crashes, possibly because I cannot try data large enough to cause it, or perhaps because of different compiler versions...

ADD COMMENTlink written 7 weeks ago by Peter Langfelder1.3k
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.2.0
Traffic: 223 users visited in the last hour