GSEA after DESeq2

580 views
Skip to first unread message

Jianli Zhou

unread,
Jul 8, 2021, 2:18:20 PM7/8/21
to gsea-help
Hi All, I am using GSEA after DESeq2 analysis and here are the 2 questions I have:

(1) I am using log2FoldChange to rank all the genes, because it makes more sense to me. I've seen others using -log10P value * sign of log2FoldChanges, statistic score etc to rank the genes. I am just wondering if using log2FoldChange is OK with GSEA.

(2) Attached is the plot of the enrichment scores of 3 pathways. The red one has the highest Normalized Enrichment Score (NES); the blue one has the lowest NES. The p-values/FDR of the 3 pathways are all below 0.05. Based on the graph, can I draw the following conclusion: the red pathway is upregulated, the blue pathway is downregulated, and the green pathway is slightly upregulated.




image (1).png

Anthony Castanza

unread,
Jul 8, 2021, 9:12:50 PM7/8/21
to gsea...@googlegroups.com
Hello,

Yes, I think Log2(FC) is a reasonable metric for GSEA preranked, it's the general "default" for most use cases.
Your interpretation of the plots is correct as well.

One additional option for a DESeq2 based workflow is to export the normalized counts table, then this can be used with standard GSEA instead of GSEA Preranked, but the only major advantage here would be if you have more than 6 samples per phenotype, which would allow you to use the phenotype permutation null distribution generation mode rather than the "gene_set" mode which is used for Preranked.

--
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/097ee131-4d03-4eaf-aa68-bbce5596dd8an%40googlegroups.com.


--
-Anthony

Anthony S. Castanza, PhD
Curator, Molecular Signatures Database
Mesirov Lab, Department of Medicine
University of California, San Diego

Jianli Zhou

unread,
Jul 8, 2021, 9:32:55 PM7/8/21
to gsea-help
Thank you very much for your help! What if the "phenotype" is a continuous variable?  

Anthony Castanza

unread,
Jul 8, 2021, 9:40:08 PM7/8/21
to gsea...@googlegroups.com

Well, I've never tried DESeq2 with continuous variables, but if you're able to export the normalized counts table output from it then GSEA also supports continuous variables through the "continuous cls" format (https://software.broadinstitute.org/cancer/software/gsea/wiki/index.php/Data_formats#CLS:_Continuous_.28e.g_time-series_or_gene_profile.29_file_format_.28.2A.cls.29)

and the use of Pearson correlation between the gene expression profiles and the time course.

 

-Anthony

 

Anthony S. Castanza, PhD

Curator, Molecular Signatures Database

Mesirov Lab, Department of Medicine

University of California, San Diego

 

Jianli Zhou

unread,
Jul 8, 2021, 9:45:03 PM7/8/21
to gsea...@googlegroups.com
Thanks a lot!

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/WG4ijQx8k38/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/SJ0PR05MB76096D96CD906ADB7BF8224EF7189%40SJ0PR05MB7609.namprd05.prod.outlook.com.
Reply all
Reply to author
Forward
0 new messages