Hello all, I am working on microarray data from GEO (GSE15919). The expression values given in the series matrix file are log2(Cy5/Cy3). I use its platform data (GPL1528) a R package "Annotationdbi" to annotate probes to be an entrez id and found that some genes are mapped by multiple probes. Is it possible to combine or average expression values from these probes? and if so, how to do this?

Thank you very much.

I almost always advise people not to average the probes but, if you want to:

library(limma)
help("avereps")

Personally I prefer to either simply keep all the probes as they are or, if reducing to one row per gene is necessary, then I keep only the best or most expressed probe for each gene, i.e., keep the probe with the highest average log-intensity over all samples.