Hi,

I am following different strategies while clustering I don't know if I performed properly the annotation while using clustering with external information. As I need to specify the regions to be annotated I have used the predefined object "promoter" as follows:

require(TxDb.Mmusculus.UCSC.mm9.knownGene)
promoter<-promoters(TxDb.Mmusculus.UCSC.mm9.knownGene)
head(promoter)

And then while annotating:

require(org.Mm.eg.db)
anno<- detailRanges(promoter, txdb=TxDb.Mmusculus.UCSC.mm9.knownGene,
                    orgdb=org.Mm.eg.db, promoter=c(3000,1000), dist=5000)
head(anno$overlap)

Is it ok? I think it is , but I got this message:

Warning message:
In detailRanges(promoter, txdb = TxDb.Mmusculus.UCSC.mm9.knownGene,  :
  strandedness in incoming regions is ignored when overlapping

Thanks!!

Sergio

Okay, you've got a lot of questions here, but let's see what we can do. For your first question, the warning message just means that the strand of the incoming regions is ignored when computing overlaps with annotated features, i.e., the function won't limit overlaps of forward-strand incoming to forward-strand annotation. If you do want this behaviour, you can get it by setting ignore.strand=FALSE in the devel version of csaw. That said, I'm not sure it makes any sense to annotate promoters, given that you already know the gene corresponding to each promoter.

As for the BED file, you're not showing what the error is (if there is any), so I can't be sure what you're doing wrong. From looking at your code, the most obvious mistake is that you should be assigning

test_up$score<- tabpromoter$logFC.up[up]

as test_up is subsetted but tabpromoter is not. There may also be an issue with using the logFC.up count as the score, but that would depend on how GREAT uses the score information.

I have also problems to try to define a bed file for GO analysis (using GREAT) for up and down regions, as the table that I get has logFC.up and logFC.down I was trying to define something like this... but I am doing something wrong obviously:

up<-tabpromoter$logFC.up>1
require(rtracklayer)
test_up<-promoter[up]
test_up$score<- tabpromoter$logFC.up
names(test_up)<-paste0("region", 1:sum(up))
export(test_up,"tabpromoter_UP.bed")

Help! XD 

Thanks!

Sergio

Hi Aaron for the answers. Concerning my second question:

tabpromoter is setted as follows:

olap<- findOverlaps(promoter, rowRanges(filtered))
olap
tabpromoter<- combineOverlaps(olap, results.promoter$table)
head(tabpromoter[!is.na(tabpromoter$PValue),])

Concerning the error:

> test_up<-promoter[up]
Error in NSBS(i, x, exact = exact, upperBoundIsStrict = !allow.append) :
  subscript contains NAs

It is maybe easier to explain what I want to do. I want to have the DBS found in promoters for several histone marks. It is true that it doesn't matter to annotate as I am using a package of annotated promoters, good point . The only thing that I need is to find a way to merge the windows on the promoters and sort those that are GAIN or LOSS because the treatment into different tables or bed files.  I have filtered using predefined regions and abundance , as follows:

ab <- aveLogCPM(asDGEList(data))
keep.ab <- ab > 0
keep.reg <- overlapsAny(rowRanges(data), regions)
keep <- keep.ab & keep.reg
filtered.data <- data[keep,]

I know..I have read plenty of times the documentation , but I do not know what it is better if doing "clustering with external information" (6.3) or merging using the tol option (6.4).  I am a bit lost-frustated. Sorry. Need just some hints I guess