using goProfiles (BioC) with non model organisms

Question

[dago]

Dear all,

During the last days I have been trying to figure out the best tool to perform gene enrichments and enrichment comparisons using GO.

It seems that the Bioconductor package goProfiles does exactly what I need. The problem is that I do not have a model organism; therefore I have no organism and annotation package available.

Looking in the manual of goProfiles, in the section "3.2.1 How can data be entered", I understand that for case like mine data can be entered as simple data frame, where are reported "GeneID", "Onology", "Evidence" (optional), "GOID".

> test.profiles
    GeneID Ontology Evidence       GOID
1 11589343       CC      IEA GO:0000156
2 11589345       CC      IEA GO:0003677
3 11588732       CC      IEA GO:0000179
4 11590726       CC      IEA GO:0003333
5 11591703       CC      IEA GO:0000156
6 11591901       CC      IEA GO:0003677
7 11591045       CC      IEA GO:0000156
8 11589721       CC      IEA GO:0000049

I tried to submit to the function basicProfile the data in such a format but I still get the same error telling me that the organism package has to be provided:

> basicProfile(test.profiles, onto = "CC", level=2)
Error in as.GOTerms.list(genelist, idType, orgPackage, anotPackage, na.rm = na.rm) : 
  'orgPackage' cannot be null. An organism annotation package (org.Xx.eg.db) must be provided

I guess I am missing something here. Is there anyone who could help me to understand my mistake?

[Devon Ryan]

As it says, it needs an annotation for your organism so it can determine the background probabilities. What are the GeneIDs from? If these are IDs from a related organism then you might be able to use the annotation for it.

[dago]

I create my own organism package with AnnotationForge. But what it is needed for goProfiles is the dataset org.Xx.egGO, which is not created w AnnotationForge. Any idea where I could look up a way to create this dataset and include it in my organism package?

[Devon Ryan]

Presumably you could use the same annotation. I'm assuming that the portion you're using now is just comprised of differentially expressed features.

[dago]

Thanks for your answer.

I can see that:

> colnames(org.Hs.egGO)
[1] "gene_id"  "go_id"    "Evidence" "Ontology"

Therefore, I assume that this dataset is a dataframe containing the above information. My problem is that I do not know which procedure to follow to include this datraframe as org.Xx.egGO in the package of my oprganism, which I created with makeOrgPackage() from AnnotationForge

[Devon Ryan]

It's not a data.frame, it just has a colnames accessor. That's a Go3AnnDbBimap object. I've never had to make one myself. Realistically speaking, it might be easier to just use a different package that lets you manually supply a background set of IDs.

[cpcantalapiedra]

I have the same problem. Did you find a solution?

Answer 1

[cpcantalapiedra]

I made it work with the idType parameter in basicProfile(), like:

basicProfile(GOTermsFrame, idType = "GOTermsFrame", onto = "BP", level = 2)

[Maria Scheel]

Hey cpcantalapiedra!

I tried using your advice, as I also use three non-model organisms with already several annotated GO terms per Gene ID - and my aim is to compare all three organisms with each other.

If I create the basic profile the way you supposed, I got 0 frequencies - did you have that as well? In which format did you upload your data initially before building the functional profile? I feel like my tables just might not have the right format, which is:

GeneID Ontology Evidence       GOID 
1 110826457       MF      IEA GO:0004888

But even if I calculated the frequencies (occurrence of GO term in data frame Cafin) manually, creating the mergedProfileList results in the error:

Error in !emptyCats : invalid argument type

Can anybody help me?