Use multiple GO terms or just one?

[aspencoyle]

Hi all,

I've been poking around the Tanner crab transcriptome, and I noticed something.

Previously, I'd been examining genes with the GO term for "immune response" (n = 87). However, substantially more genes have the GO term for "innate immune response" (n = 428).

I've done quite a bit of research on the 87 immune response genes, so I'd prefer not to throw those out. However, it really seems appropriate to include those 400-odd other immune-related genes.

I know that generally, grouping together genes from a whole bunch of different GO terms is frowned upon. But in this case, would it be acceptable for my methods section to include something like

The cross-referenced table of accession IDs and GO terms for each transcriptome was filtered to examine genes with the GO term for "Immune Response" (GO: 0006955) and "Innate Immune Response" (GO:0045087). Literature searches for function in closely related species were then performed

Or should I restrict myself to only including genes matching a single specific GO term?

Thank you!

[kubu4]

I don't have a response to your question, but just wanted to clarify this:

I know that generally, grouping together genes from a whole bunch of different GO terms is frowned upon.

This isn't frowned upon, as this is essentially what a GOslims do (group related GO terms into a higher level group), which in turn ends up grouping the genes.

[aspencoyle]

Sounds good - I suppose what I mean is grouping them together more arbitrarily isn't acceptable methodology for a paper (such as, say, including all GO terms with "immune" somewhere in the name)

[kubu4]

Well, as it turns out, that is also acceptable. That type of grouping is called "semantic similarity", where GO terms a grouped based off of how the words in a GO terms are related.

[aspencoyle]

Well holy moly, thanks a bunch - good to know!

[sr320]

Previously, I'd been examining genes with the GO term for "immune response" (n = 87). However, substantially more genes have the GO term for "innate immune response" (n = 428).

I presume this does not involve the discussion of genes determined to be influenced by temperature or time as we searched for the word "immune"?

[aspencoyle]

Nope, this is for the general description of immune-related genes within the transcriptome

[sr320]

So finish up differential gene expression aspect first.. then we can move back to address general description as this is secondary.

[aspencoyle]

Alright, think I've finished a pretty good description of differential gene expression (or made some substantial progress - some room left for improvement in the parasite differential expression analysis, and a round of edits needed throughout, but the gist is there)

[sr320]

As a more general comment here, it is important to remember GO is hierarchal and it is good idea to leverage ancestor/parent (GOslim) terms to capture relevant genes