High Resolution images for Pathway Enrichment analysis

[Rober Abdo]

Hello GSEA group,

I wonder how I would download high resolution-publishable images of pathway analysis from the GSEA software.

With appreciation,

Rober

[Anthony Castanza]

Hi Rober,

 

By default GSEA writes out all the images it produces to a directory on your system, the GSEA output directory can be located by clicked the GSEA File menu, then Preferences, then the directory is listed under Report Settings Default Output folder. Within that directory will be a directory for each date and then each run of GSEA has a directory within that.

 

If the images in that folder aren't high quality enough, GSEA does support the ability to produce SVG plots. You can enable this in the GSEA software by expanding "Advanced Fields" then then setting "Create SVG plot images" to "true". You would want to then rerun GSEA to create a result folder with these high resolution images.

 

It is important to note however, that when re-running GSEA you will see some variance in the result as a consequence of the random number seed used by default to generate the permutation matrix for null distribution testing. If you want to reproduce the previous run exactly, the random seed value should be noted in the "Comments" section on the index.html report page. You can paste that value into the "Seed for permutation" parameter also under "Advanced fields" replacing the default "timestamp" value.

 

Hope this helps, let me know if you have any other questions

 

-Anthony

 

Anthony S. Castanza, PhD

Curator, Molecular Signatures Database

Mesirov Lab, Department of Medicine

University of California, San Diego

--
You received this message because you are subscribed to the Google Groups "gsea-help" group.
To unsubscribe from this group and stop receiving emails from it, send an email to gsea-help+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/gsea-help/791499bc-0632-4bdb-84f7-11a44dc24600n%40googlegroups.com.

[Rober Abdo]

Thank you very much, Anthony, for the prompt-productive response. This is very helpful.

With appreciation,

Rober

To view this discussion on the web visit https://groups.google.com/d/msgid/gsea-help/SJ0PR05MB76095D70CA81D3722629520BF7519%40SJ0PR05MB7609.namprd05.prod.outlook.com.

[Rober Abdo]

Hello Anthony,

I have enabled the SVG plot feature by indicating true in the field, but I still do not get HD images in the result folder; they are PNG images, but they lack clarity upon zooming in a little.

Thanks,

Rober

[Anthony Castanza]

Hi Rober,

 

I just ran a quick test to double check that there aren't any issues producing the plots on our end and it definitely should be working.

The svg images are only located in the job results directory and aren't used in the HTML results webpage. They are also gzipped by default, so you'll want to go into the specific report directory (make sure it's directory for the new run and not the old run) and then look for enplot_NAME_OF_GENESET_[#].svg.gz

 

An example of the kind of thing you'd be looking for:

-Anthony

 

Anthony S. Castanza, PhD

Curator, Molecular Signatures Database

Mesirov Lab, Department of Medicine

University of California, San Diego

To view this discussion on the web visit https://groups.google.com/d/msgid/gsea-help/7459d760-a8c8-4be2-8622-67b3701df105n%40googlegroups.com.

[Rober Abdo]

Thank you very much, Anthony, for the detailed clarification.

Rober

To view this discussion on the web visit https://groups.google.com/d/msgid/gsea-help/CAGCeyZwjwEFUk_37waFXPuTZVrYVj4vPY7ci5t_U7KoYpJ1u3A%40mail.gmail.com.

[Rober Abdo]

Hello GSEA,

I am getting an FDR and p- value of (0) for the top enriched pathways although I have used the whole gene expression matrix for the two groups, not the DEGs. Is such (0) value realistic?

Thanks,

Rober

[Anthony Castanza]

Hi Rober,

 

Yes, GSEA generates an empirical pValue that is, a pValue derived directly from the permutation matrix, so if of your 1000 permutation, 500 of them have an enrichment score for that set of the same sign as the true enrichment score, then the pvalue is Number of scores >= the true score/the number of scores of the same sign. If an enrichment as strong as the true enrichment was never observed in the distribution of permutation derived enrichment scores then the result is a pValue of exactly 0.

 

Does that make sense?

 

-Anthony

 

Anthony S. Castanza, PhD

Curator, Molecular Signatures Database

Mesirov Lab, Department of Medicine

University of California, San Diego

 

From: gsea...@googlegroups.com <gsea...@googlegroups.com> on behalf of Rober Abdo <rober...@gmail.com>
Date: Tuesday, February 8, 2022 at 2:43 PM
To: gsea...@googlegroups.com <gsea...@googlegroups.com>
Subject: [gsea-help] FDR and p-value of (0)

Hello GSEA,

 

I am getting an FDR and p- value of (0) for the top enriched pathways although I have used the whole gene expression matrix for the two groups, not the DEGs. Is such (0) value realistic?

 

Thanks,

 

Rober

 

On Thu., Jan. 13, 2022, 09:23 Rober Abdo, <rober...@gmail.com> wrote:

Thank you very much, Anthony, for the detailed clarification.

 

Rober

 

On Wed, 12 Jan 2022 at 14:00, Anthony Castanza <acas...@cloud.ucsd.edu> wrote:

Hi Rober,

 

I just ran a quick test to double check that there aren't any issues producing the plots on our end and it definitely should be working.

The svg images are only located in the job results directory and aren't used in the HTML results webpage. They are also gzipped by default, so you'll want to go into the specific report directory (make sure it's directory for the new run and not the old run) and then look for enplot_NAME_OF_GENESET_[#].svg.gz

 

An example of the kind of thing you'd be looking for:

To view this discussion on the web visit https://groups.google.com/d/msgid/gsea-help/CALTkV2mA_S2KBghj0mdfVwrea2Br-TvOJ3Qu2-J8x5bGAnUMGg%40mail.gmail.com.

[Rober Abdo]

Hello Anthony,

Thank you for the clarification.

Would you recommend me here to increase the number of permutation? Initially, I had it as you indicated (1000).

With appreciation,

Rober

To view this discussion on the web visit https://groups.google.com/d/msgid/gsea-help/SJ0PR05MB7609FFB76B17DA7FF0241469F72D9%40SJ0PR05MB7609.namprd05.prod.outlook.com.

[Anthony Castanza]

You can if you want to, assuming you have enough samples to support a larger permutation number (if running in Phenotype permutation mode). If running in Gene Set permutation mode/Preranked, a typical experiment generally has enough genes to support tens of thousands of gene set permutations, but other than the possibility of getting a non-strictly-zero answer there wouldn't be much of a benefit to doing so, and it would substantially increase the computation time.

 

You would also experience some of the minor variability that can occur from run-to-run as a result of the random seed used to initialize the permutation calculations, although that can be prevented by reusing the seed from the initial run (it's printed under comments on the index.html page from the initial run, and can be provided to the seed for permutation field under Advanced fields).

To view this discussion on the web visit https://groups.google.com/d/msgid/gsea-help/CALTkV2mxisyOH_vu5pE-rDvVK5uz0MoqHR03Kq7zE3ru7TbWFA%40mail.gmail.com.

[Rober Abdo]

Thank you very much for the prompt and detailed response.

Rober

To view this discussion on the web visit https://groups.google.com/d/msgid/gsea-help/SJ0PR05MB760996A4E0BCF6FE90A9EF2EF72E9%40SJ0PR05MB7609.namprd05.prod.outlook.com.