This is a test version of Biostars. For the public version, visit https://www.biostars.org.
Gene set enrichment analysis input data

I apologize for asking this naive question but I have read several tutorials and I feel unsure about the following: When using gene set enrichment analyses ( ReactomePA, clusterProfiler, gprofiler2 etc.) The input data is an order ranked geneList

  1. Is this geneList the output from DESeq2 or a filtered list for example filtered by FDC or log2FoldChange?

  2. This is how most tutorials filter the gene set in R:

    gene <- names(geneList)[abs(geneList) > 2]

Is this assigning the gene variable names of genes with values greater than the absolute value of 2?

Thanks!

gse deseq2
• 2,763 views • link
written by ATCG ATCG

These tools perform enrichment analysis for a list of selected genes, not gene set enrichment analysis. Edit: Some of these tools do not perform GSEA as mentioned below. GSEA would be comparing the entire set of all your genes ranked by e.g. _´log10(pvalue)` against a a priori defined set of genes, e.g. genes contributing to a pathway.

In your case one typically selects genes below a certain padj cutoff, e.g. 0.05. I prefer to divide into up- and downregulated genes to make interpretation more straight-forward.

• 0 views • link
written by ATpoint ATpoint

Are you sure about this?

gsePathway {ReactomePA} R Documentation

gsePathway Description Gene Set Enrichment Analysis of Reactome Pathway

gseKEGG {clusterProfiler} R Documentation

gseKEGG Description Gene Set Enrichment Analysis of KEGG

• 0 views • link
written by ATCG ATCG

1 answer

clusterProfiler does perform GSEA. For the kegg enrichment analysis, you start with a list of pathways or kegg genes (KOs) in your sample of interest compared to a background.

https://yulab-smu.github.io/clusterProfiler-book/chapter6.html#kegg-over-representation-test

• 0 views • link
written by N15 N15

I see, edited my comment, thanks.

• 0 views • link
written by ATpoint ATpoint

Log in to answer this question.

More posts like this