Too Many Gene Sets Rejected After Restricting to Dataset

[Daniel Wang]

Hello,

I ran GSEA v4.1.0 on a list of 426 genes using their normalized and log2-transformed read count data, with phenotype labels of response and nonresponse to anti-PD-1 immunotherapy. After using the H, C2, C6, and C7 gene sets and the Ensembl CHIP file to run GSEA, I found that the vast majority of the gene sets were rejected and filtered out after “restricting to the dataset,” even after I changed the gene set size filters to have a min of 10 and max of 5000. For example, 11155 out of the 11340 total gene sets were filtered out, and 49 of the 50 hallmark gene sets were rejected. Is this normal, and if not, how can I fix it? The genes used had an adjusted p-value of less than 0.05 from differential expression analysis with DESeq2. Any help is appreciated.

Thank you,

Daniel Wang

[Anthony Castanza]

Hi Daniel,

 

GSEA is intended to be run on the entire expressed gene list, not a list restricted to, for example, just significantly differentially expressed genes. We also recommend normalized counts that haven’t been log transformed as the input, but the log transformation is just a matter of not compressing the range of the signal to noise distribution and isn’t nearly as important as being sure to supply the information for all of the expressed genes.

 

-Anthony

 

Anthony S. Castanza, PhD

Curator, Molecular Signatures Database

Mesirov Lab, Department of Medicine

University of California, San Diego

http://gsea-msigdb.org/

--
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/66f78408-dce0-4ffd-a5e8-f106a868729fn%40googlegroups.com.