Negative ES Interpretation

[holland...@gmail.com]

Hi all,

I hope everyone is doing well in the pandemic. Keep safe out there.

I have a question about how to interpret negative ES values. Here's a bit of context: I want to test the hypothesis that a given pathway is enriched in one particular tissue type relative to another tissue type. To perform this test I conducted two separate experiments using RNAseq data, and I wanted to check if the same pathway was enriched in both experiments. These experiments are replicates of each other. I obtained normalized read count data using TMM. After running GSEA for the pathway that I am interested I was surprised to realize that one experiment showed a positive ES value, while the second showed a negative ES value. As a consequence of this,one of the outputs says that the pathway is upregulated for condition #1, while the second output flags condition #2 as being upregulated.

When reading the GSEA documentation and the GSEA manuscript it is stated that: "A positive ES indicates gene set enrichment at the top of the ranked list; a negative ES indicates gene set enrichment at the bottom of the ranked list." An important piece of information here is that the vast majority of the RNA sequenced genes are positively correlated with one of the two phenotypes. The zero cross depicted in the ranked list of genes in Figure 1 from the 'Snapshot' of the results is at around 17k genes when I have 18k genes in total. In both of the experiments, all of the genes from the pathway that I am interested in are depicted as being ranked before the zero cross, and thus always correlated with condition #1. My interpretation of this results would be that even though there is a difference regarding where the peak of the ES is (either at the 'top' or the 'bottom' of the list), both analyses are saying that the ES is significantly enriched for condition #1, since all of the genes in the pathway are assigned to ranks before the 'Zero' cross.

Would this be correct? I'm feeling a bit thrown off by this apparent conflict between what the description of the ES value says and what the output is calling as being 'enriched.'

Thanks so much for your time.

Best,

P

[Anthony Castanza]

Hi,

Assuming I understood things correctly; that Condition #1 reflects the same biological condition in both experiments, and that the CLS files have been designed in such a way that you've selected condition 1 vs. condition 2 for both experiments and not condition 2 vs condition 1 in the second experiment, then this is quite concerning.

A positive enrichment score should always reflect enrichment on the positive side of the zero cross (although not necessarily all genes on the positive side) and be enrichment in whichever phenotype was selected to be first in the comparison in the Phenotype selection dialogue. And vice versa, with a negative enrichment score reflecting enrichment in the genes on the negative side of the ranked list.

GSEA is known to have some issues with highly skewed gene distributions, but that shouldn't affect the raw enrichment scores, just NES and significance calculations when GSEA runs in to a condition where its only sampled from one side of the distribution.

Would you be willing to share your data (confidentially!) so we can look in to this further? You can send datasets to gsea...@broadinstitute.org

We'd at least like to see the two gene set results you're specifically referring to if you're not, you can always black out anything confidential on them if you'd prefer not to share more data.

-Anthony

 

Anthony S. Castanza, PhD

Curator, Molecular Signatures Database

Mesirov Lab, Department of Medicine

University of California, San Diego

[holland...@gmail.com]

Hi Anthony!

Thanks for answering so quickly. I appreciate it.

I believe I have set up the analysis properly, yes. It was one of the first thing I checked, and I ran and re-ran the analyses just to be sure that I had not screwed anything up.

I will shoot you an e-mail with the data. I would appreciate if it could be kept confidential and we can come back to this chat once we figure out what happened.

Thanks,

Best,

P

[Pietro Mello]

Hey, Anthony

I sent you my files yesterday about the negative ER values in GSEA. Could you tell me if you got them alright? I sent them via my @ku.edu e-mail (holland...@ku.edu).

Best,

P

--
You received this message because you are subscribed to a topic in the Google Groups "gsea-help" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/gsea-help/RN5f-dTzwzY/unsubscribe.
To unsubscribe from this group and all its topics, send an email to gsea-help+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/gsea-help/489a4b52-eb0d-4aa0-ab73-c6f7412de4bf%40googlegroups.com.

--

Pietro Longo Hollanda de Mello

http://lattes.cnpq.br/2395634311539356

Doctoral Candidate, Department of Ecology and Evolutionary Biology

University of Kansas

Dyche Hall, 1345, Jayhawk Blvd

Lawrence, KS

E-mail: holland...@ku.edu

[Anthony Castanza]

Hi,

 

No, I did not receive any files. You’ll probably need to resend them. Just in case, the address is gsea...@broadinstitute.org.

It might be best to out them in a google drive folder or similar and just share a link to them that way.

--
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/CACwjVxGEeFGe%2BoXDMXCXi3kSgc6VKsfhBbwsDdfm4KOpTCgEvA%40mail.gmail.com.

[Pietro Mello]

Gotcha,

I'll do it from this e-mail address them. Might be easier. I got the auto-reply when I sent it, but since you had replied so quickly the first time I figured something could have gone wrong. I'll send it to the e-mail address you provided and I'll shoot you an e-mail through here so that you know it's been sent out.

P

To view this discussion on the web visit https://groups.google.com/d/msgid/gsea-help/BYAPR05MB5782D5A33386D7D05FC20AA2F7BF0%40BYAPR05MB5782.namprd05.prod.outlook.com.

[Pietro Mello]

Sent it your way with attachments in Google Drive.

Also got the auto-confirmation e-mail.

Please let me know if everything looks good.

Thanks much.

Best,

P

[Anthony Castanza]

Hi,

 

It came through this time, and one of my colleagues forwarded me the original email which apparently came through for everyone but me. I’ll take a look at these files as soon as possible and get back to you.

 

-Anthony

To view this discussion on the web visit https://groups.google.com/d/msgid/gsea-help/CACwjVxEeG8XebjMU8yo69qiHhFAHVMZw2DcQZKoLu7RSzWbLZw%40mail.gmail.com.