GO enrichment on up and down regulated genes on treated sample

[kate2]

I am new to RNA-sequencing analysis and have just successfully run my first analysis on differential gene expression ( ).

I was then keen to get a picture of what was up and down regulated in the treated sample so, as a second and third run on the data, I extracted the genes where log2(FC) >= 0, and then where log2(FC) <=0.

Am I correct in thinking that this has given me the GO analysis on what was up-regulated in the treated sample (factor level 1) and then what was down regulated in the treated sample respectively compared to the untreated (factor level 2).

Or is there a better way of doing this?

Happy to post my history if this all appears nonsense.

[jennaj]

Hi @kate2

Congrats on learning a new analysis process!

We have a few tutorials that explain more about how the inputs and queries for these tools work.

Generating count tables

• Hands-on: 1: RNA-Seq reads to counts / 1: RNA-Seq reads to counts / Transcriptomics

DE with Limma

• Hands-on: 2: RNA-seq counts to genes / 2: RNA-seq counts to genes / Transcriptomics

GOSeq → GOEnrichment

• Hands-on: 3: RNA-seq genes to pathways / 3: RNA-seq genes to pathways / Transcriptomics

GOEnrichment details

• Hands-on: GO Enrichment Analysis / GO Enrichment Analysis / Transcriptomics

---

With that context, I think you are asking about making a direct query against the outputs from a tool like Limma, correct?

@kate2

I was then keen to get a picture of what was up and down regulated in the treated sample so, as a second and third run on the data, I extracted the genes where log2(FC) >= 0, and then where log2(FC) <=0.

If your DE table was filtered already, the result describes the first factor in the presence of the second factor. For a query against the fold change between the two samples: a positive value means overexpressed and negative means underexpressed.

You could explore that result more with a tool like this one:

• Hands-on: 2: RNA-seq counts to genes / 2: RNA-seq counts to genes / Transcriptomics (MD plots and Volcano plots)

Then see the next step in the tutorial above-- the data is further reduced by a query against the fold change and a pvalue. These seems to be what you want to do with your direct queries here:

@kate2

Am I correct in thinking that this has given me the GO analysis on what was up-regulated in the treated sample (factor level 1) and then what was down regulated in the treated sample respectively compared to the untreated (factor level 2).

After, once you have isolated your significant genes of interest, and annotated them with true/false (required for GOSeq), you can then explore with GOSeq and GOEnrichment using the reduced data.

In short: the processing has major steps where the data is reduced. Generating the counts is one of those, then the next reduction step comes from determining the significant differences between your samples. You can do this a few ways! Then, once you have a set of genes of interest, you can explore the connections between them using higher levels of annotation (GO → Pathways → biological meaning).

Please give that a review and let us know if it answers your question! We haven’t summarized this content at the forum in a while, so I’m glad you asked!