Gene- Level Go Enrichment Analysis- with annotation on transcript-level

Question

[biosjm]

Hello, I have a question regarding GO enrichment analysis: I`ve identified DEG by Deseq2 and would like to perform a GO enrichment analysis with topgo or similar for this set of data. I have an annotated transcriptome from Trinotate, but it contains the go term information for several transcripts from one gene. According to the vignette of Topgo I have to prepare a list o the gene id and respective go terms. My question are now:

a) Do I map now all the go terms of all transcripts from one gene together?

b) Do I have to de-duplicate, if always the same go terms occur between the transcripts within one gene?

gene / transcripts / GO

1 / 1a / GO:123, GO:456, GO:987

1 / 1b / GO:123, GO: 000

1 / 1c / GO:123, GO: 456

--> prepare list (in case of b) Is this right?

gene / GO

1 / GO:123, GO:456, GO: 987, GO:000

Thank you for any comment or link to clarify this!

Answer 1

[Jean-Karim Heriche]

a) Yes unless you have reasons to want to distinguish between different products of a gene.
b) It can't hurt to ensure there are no duplicates (it probably doesn't matter but without looking at the implementation, I can't be sure). You can't have duplicated genes.

[biosjm]

Thank you for your answer! I agree with you that it might not be a good idea to have duplicated genes as input for topgo. I was wondering, if I have to keep duplicated go terms for one gene : gene / GO 1 / GO:123, GO:456, GO:987, GO:123, GO: 000, GO:123, GO: 456

or I have to de duplicate the terms as above in b)

[Jean-Karim Heriche]

To clarify, my answer to b was that I don't know if duplicated GO terms are an issue but that removing duplicates doesn't cost you anything. I added the warning about duplicated genes because not only is it not a good idea, topGO would throw an error.

[biosjm]

Thank you for your help!