This is a test version of Biostars. For the public version, visit https://www.biostars.org.
GSEA and enrichment analysis of sequence data

I have some sequence data which are not expression. Is it possible to do GSEA or GO enrichment?

My understand:If I want to perform GSEA, I need expression level data.

My data are like this:

  1. A set of pair regions which interact with each other (RNA-DNA);
  2. CAGE-seq data (several regions which has "cap", probably they are TSSs or very close to enhancers or promoters).

I tried CAGE-seq (mainly) with GREAT and chipenrich (used as ChIP-seq). I think all of these methods focused on the areas close to the regions I have. So I think it could be a little dangerous because sometimes enhancers will not only regulate areas right near? The results may not be very accurate?

So my questions are:

  1. For non-coding regions, are there any standard workflow for GSEA and GO enrichment?
  2. If I use the interacted regions to do with GREAT or chipenrich, will this be reliable? Because I think 2 regions if they have interaction, the GREAT analysis or chipenrich analysis based on these regions are much reliable.
go-enrichment non-coding-regions gsea
• 1,339 views • link
written by Bot_COLA • Bot_COLA

1 answer

Technically speaking, gene set analysis only requires having a statistical value for each gene. This could be a signed statistic like a T value, or an unsigned statistic like a p-value.

For CAGE-seq, it would be fair to assign statistics for a CAGE peak to the gene with the closest TSS if that TSS is within 1kb.

I'm not ceratain what an "RNA region" is in the context of "RNA-DNA" interaction. Is this something like a correlation between two CAGE regions (one TSS-proximal and likely a promoter RNA, and one TSS-distal and likely an enhancer eRNA?) Or is it something like a correlation between ATAC (open chromatin) between a promoter and a distal peak? Or something else?

• 0 views • link
written by LChart LChart

Sorry for late reply, and thank you so much for comment!

I am using assign statistics for CAGE peak right now. I think this could work

About interaction, your understand is correct but this interaction also includes the "physically close". Probably they don't have interaction, they were captured just because they are very close at that time. So I may choose like promoter-enhancer.

• 0 views • link
written by Bot_COLA • Bot_COLA

Log in to answer this question.

More posts like this