topGO approach for a list a genes without expression

Question

[serpalma.v]

Hello,

I would like to know if this approach is correct. Briefly, I have a list of targets for a miR and I want to perform GO unsing topGO. TopGO needs a function (geneSel) telling it which genes to choose from the gene set, which is usually the differentially expressed ones. Since I have a list of genes (targets for the miR) I have no expression values and therefore, no sigincancies. So, this is what I did, but I am not sure if is correct:

1. get all gene names:

   mart <- useDataset("hsapiens_gene_ensembl", mart=useMart("ensembl"))
   all_ensembl_gene_id <- getBM(attributes = "ensembl_gene_id",
                                values = "*", mart = mart)
   

2. Select the targets IDs and give them value of 1. The rest of IDs in the set are 0s.

   all <- factor(as.integer (all_ensembl_gene_id[,1] %in% targets))
   names(all) <- all_ensembl_gene_id[,1]
   

3. Perform gene ontology, choosing only targets (1's)

   GOdata <- new("topGOdata",ontology = "CC",
              allGenes = all,
              geneSel=function(p) p == 1,
              description ="miR-targets",
              annot=annFUN.org, mapping="org.Hs.eg.db", ID="Ensembl")
   

Do you think this is a valid way to perform a GO. I am fairly new to this and I would appriciate your help very much.

Thanks

Answer 1

[Sudeep]

I do not see any problems as such with your R code, but instead of using the function, you can also use a vector of gene ids as inputs. Have a look at this topGO vignette esp. sections 4.4 through 4.6

[serpalma.v]

Thank you for replying.

The vector of gene ids would be my targets, right?

and, do you think that using all gene IDs as background is a valid approach?

Cheers!