PeptideShaker for HPC cluster

[HFaz]

Hi All,

Is there any tutorial for running PeptideShaker on HPC cluster? 

Thanks

Hfaz

[Marc Vaudel]

Hi,

You will find the documentation on how to run our tools in command line in our wikis:

https://github.com/compomics/searchgui/wiki/SearchCLI

http://compomics.github.io/peptide-shaker/wiki/peptideshakercli.html

For the setup of clusters there are a few things we recommend:

- keep a "clean" copy of the tool and distribute it before execution. Don't run multiple command lines from the same instance.

- use the 'temp_folder' option to locate the temporary files of each instance

- use the 'zip' and 'data' options to package the results

Hope this helps, don't hesitate to write again in case anything is unclear :)

Best regards,

Marc

--
You received this message because you are subscribed to the Google Groups "PeptideShaker" group.
To unsubscribe from this group and stop receiving emails from it, send an email to peptide-shake...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Felix Boos]

Hi,
I have two additional questions about running SearchGUI and PeptideShaker on a HPC cluster:
- do they require OpenGL?
- how many cores can they use, i.e. is there a limit as of which more cores will not be beneficial?
Thanks,
Felix

[Harald Barsnes]

Hi Felix,

Sorry for the late reply. The answers are quite simple though. No, OpenGL is not required. And no, there is no generally limit on the number of cores that can be used.

Regarding a limit on when adding more cores will not be beneficial we have not done any such tests ourselves. But it would very much depend on the process. For example the search engines will most likely utilize the cores differently. So the only way to know would be to test I'm afraid. But we would be very interested to know how it turns out.

Best regards,

Harald

[Innocent Onsongo]

Hi, 

I have a question about running PeptideShaker on an HPC cluster. I have a script that runs SearchGUI in parallel. Each SearchGUI instance uses a different search engine. This produces two different outputs for each search engine. My question is can I also run PeptideShaker separately for each search engine results or do I need to combine SearchGUI output? I basically want to know if I can parallelize PeptideShaker to use multiple nodes on a HPC cluster. 

Thanks!

To unsubscribe from this group and stop receiving emails from it, send an email to peptide...@googlegroups.com.

[Marc Vaudel]

Hi Innocent,

After SearchGUI, PeptideShaker creates a unified set of peptide-spectrum matches. This needs to be done for all search engine results together, so I am afraid running different instances of PeptideShaker on different nodes is not possible.

Hope it helps,

Marc

To unsubscribe from this group and stop receiving emails from it, send an email to peptide-shake...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/peptide-shaker/e1f27b25-d339-4eba-8404-ffbe27b52da0%40googlegroups.com.