Failure to run COBRA on computing clusters

[Reza Yasemi]

Hello, 

I have been a COBRA user for a while now and I can successfully run FBA, FVA and dynamic versions of these methods on my personal laptop. Recently, I am working on a genome-scale model for which I want to do DFVA, and I run out of RAM. Although I could run my Matlab codes in the form of "batch" jobs both on my personal laptop computation cluster and on Compute Canada clusters, I fail to run my Matlab codes which involve COBRA submodules. I attached the full error, in short, it points to the worker's inability to read/access the COBRA submodule. 

Any ideas on how to tackle this issue are greatly appreciated. 

Thanks, 

Reza

Ph.D. candidate

Research Laboratory in Applied Metabolic Engineering
Polytechnique Montréal 

[Thomas Pfau]

Hi,

It looks like you are using some code that tries to access files on your local machine (all the C:\ folder indicators point to it). Alternatively, if you copied the cobratoolbox folder over to the cluster, you might have added some file (e.g. a startup.m) that contains a path list with pathes on your machine.
I would suggest to install the toolbox diectly on the cluster (not copying it over from your machine) and only copy the actual scripts running your analysis.

Maybe this helps,

Thomas

--

---
You received this message because you are subscribed to the Google Groups "COBRA Toolbox" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cobra-toolbo...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cobra-toolbox/a60373c2-56e2-4827-8cba-d557e7934230n%40googlegroups.com.