I've recently started trying to process data from the GEO datasets that are publically available.
Unfortunately my R skills are still quite not up to par.

I've been working with a specific set in particular GSE3933
the code is simply to extract the expression for each sample then match it with the appropriate gene symbol/annotation.

gse3933 <- getGEO('GSE3933',GSEMatrix=TRUE)
gpl <- annotation(gse3933[[3]])
platf <- getGEO(gpl, AnnotGPL=TRUE)
anot <- data.frame(attr(dataTable(platf), "table"))
Gmicro <- exprs(gse3933[[3]])

My problem arises in two parts:

Part 1
I've accessed the user submitted annotation which is GPL3289 - this has ten variables. However there is no variable which is "gene symbol". My guess is that the GB_List variable column can be converted to Gene symbols/ Gene ID's - however I have not been able to figure out how to do this.

Part 2
I then want to match the Gene Symbols/Probes with the appropriate expression values from the matrix 'Gmicro' as per the code. Is this simply matching the ID's from the matrix generated by using the exprs function to those generated using dataTable?

The GB_list column contains comma-separated GenBank IDs, which you can convert to symbol and Gene IDs using the org.Hs.eg.db package. Unfortunately there are multiple IDs per row, so you will have to decide how you want to resolve that. But in general, you use the select() function.

And note the annotation already comes with the GSE that you downloaded, and can be found in the featureData slot. But I think you will be sorely disappointed. Those GenBank IDs look like they are all IMAGE clones that will probably take quite a bit of digging on your part to annotate. As an example:

> gbs1 <- pData(featureData(gse3933[[1]]))$GB_LIST
> ind <- rep(1:length(gbs1), sapply(strsplit(as.character(gbs1), ","), length))
> gbs1 <- unlist(strsplit(as.character(gbs1), ","))
> mapped <- select(org.Hs.eg.db, gbs1, c("ENTREZID","SYMBOL"), "ACCNUM")
> apply(mapped[,2:3], 2, function(x) sum(!is.na(x)))
ENTREZID   SYMBOL
     546      546

I created the 'ind' vector so I could map things back to the original data (accounting for the replicates), but no matter; most of these are ESTs and they don't appear to have either symbols or Entrez Gene IDs (there are over 70k IDs that I tried to map, and only 546 were successful).

So that is how you would annotate, given the infrastructure in Bioconductor.

Best,

Jim