Hello all,

I am trying to subset ESet object base on feature data.  Eset with with feature data was created using following code.

library(ragene10sttranscriptcluster.db)
library(annotate)
library(R2HTML)
library(limma)

ID <- featureNames(eset)  

Symbol <- getSYMBOL(ID, "ragene10sttranscriptcluster.db")
Name <- as.character(lookUp(ID, "ragene10sttranscriptcluster.db", "GENENAME"))
tmp <- data.frame(ID=ID, Symbol=Symbol, Name=Name)  
fData(eset) <- tmp  

> eset
ExpressionSet (storageMode: lockedEnvironment)
assayData: 28826 features, 13 samples 
  element names: exprs 
protocolData
  sampleNames: M11001NB1H-Heart-1-RatGeneArray.CEL M11003NB1H-Heart-3-RatGeneArray.CEL ...
    M11024NB2H-Heart-24-2-RatGeneArray.CEL (13 total)
  varLabels: ScanDate
  varMetadata: labelDescription
phenoData
  sampleNames: M11001NB1H-Heart-1-RatGeneArray.CEL M11003NB1H-Heart-3-RatGeneArray.CEL ...
    M11024NB2H-Heart-24-2-RatGeneArray.CEL (13 total)
  varLabels: sample
  varMetadata: labelDescription
featureData
  featureNames: 10700001 10700003 ... 10940690 (28826 total)
  fvarLabels: ID Symbol Name
  fvarMetadata: labelDescription
experimentData: use 'experimentData(object)'
Annotation: ragene10stv1 

Now I want to get eset object only where Symbol is available for corresponding ID 

sub<-eset[fData(eset)[2]!="NA"]

This gives me whole 28826 features again.

I appreciate any help.

Regards,

You don't want to use getSYMBOL() any longer. Instead you want to do

gns <- select(ragene10sttranscriptcluster.db, featureNames(eset), c("SYMBOL","GENENAME"))

at which time you will get a warning stating that you have multiply-mapped probesets. You will have to deal with this as you see fit; I tend to simply keep the first of the duplicates, as in aggregate this tends to select what I consider to be the 'right' one.

gns <- gns[!duplicated(gns[,1]),]
fData(eset) <- gns

And you can also then do

sub <- eset[!is.na(gns[,2]),]