Dear all, I am using GOseq on a list of diffentially expressed genes and I would like to do an analysis at a specific GO level (level 4 for example) instead of the complete GO. Is there a tool in GOseq that already does it ? Or do I have to prepare the file for the gene2cat option by myself ? A somewhat similar question is : how to use the GOslim ontologies (a subset of the terms in the whole GO) in GOseq ? Thank you in advance. Best regards, Amandine ----- Amandine Fournier Lyon Neuroscience Research Center & Lyon Civil Hospital (France)

<amandine.fournier at="" ...=""> writes: > > Dear all, > > I am using GOseq on a list of diffentially expressed genes and I > would like to do an analysis at a specific GO > level (level 4 for example) instead of the complete GO. > Is there a tool in GOseq that already does it ? Or do I have to > prepare the file for the gene2cat option by myself ? > A somewhat similar question is : how to use the GOslim ontologies > (a subset of the terms in the whole GO) in > GOseq ? > > Thank you in advance. > Best regards, > Amandine > > ----- > Amandine Fournier > Lyon Neuroscience Research Center > & Lyon Civil Hospital (France) Dear Amandine, Goseq doesn't let you select a specific GO level, but like you suggested, it's possible to do this manually with the gene2cat option. This would require the level of each GO term to be know. Unfortunately I can't point you towards a database with this information, but perhaps someone else on the bioconductor list knows. For GOslim, the situation is similar, in that you need to provide the gene2cat list yourself. Below is an example of how to create this using biomart. # First get the GOslim terms from biomart library("biomaRt") ensembl = useMart("ensembl",dataset="hsapiens_gene_ensembl") go_slim=getBM(attributes="goslim_goa_accession",mart=ensembl)[,1] # Now get the mapping between genes and all GO terms go=getgo(names(genes),"hg19","ensGene") # Filter the list for only GOslim terms go_slim2cat=lapply(go,function(x){ x[x %in% go_slim] }) # Run goseq with the slimmed list GO.wall=goseq(pwf,"hg19","ensGene",gene2cat=go_slim2cat) By the way, it came to our attention recently that goseq will sometimes rank the root GO terms like, GO:0003674 - Molecular Function, very high. This is to do with the way goseq handles genes with no GO annotation. It you find this is swamping your results you might like to try the developmental version of goseq, which gives an alternative way of handling genes without GO annotation. Cheers, Nadia.