# Match array probes with gene IDs from biomaRt
# efg, Stowers Institute, 3 Feb 2006

# UNIX:  change "BasePath" based on Windows/UNIX platform
switch(.Platform$OS.type,
         windows = BasePath <- "U:",
         unix    = BasePath <- "/n/projects",
         stop("unsupported OS platform")
      )

filename <- paste(BasePath, "/efg/camda/2006/arrayprobes.csv", sep="")
array.probes <- read.csv(filename, as.is=TRUE)
probe.list <- array.probes$probe
length(probe.list)
length(unique(probe.list))

GeneInfo.List <- NULL

library(biomaRt)
mart <- martConnect()

for (i in 1:length(probe.list))
{
  probe <- probe.list[i]
  cat(i, probe, "\n")  # show progress
  flush.console()

  # Assume embl ID
  GeneInfo <- getGene(id=probe,type="embl",species="hsapiens",mart=mart)
  if ( is.na(GeneInfo@table$symbol[1]) )
  {
    # IF embl ID fails, try as refseq
    GeneInfo <- getGene(id=probe,type="refseq",species="hsapiens",mart=mart)
  }

  # Can be more than one gene ID.  "unlist( lapply ..." is ugly way to fix
  for (k in 1:length(GeneInfo@table$symbol) )
  {
    GeneInfo.List <-  rbind( GeneInfo.List,
                             c(probe,
                               unlist(  lapply(GeneInfo@table, "[", k)[c(1,3,4,5,6,7,2)]) ))
  }
}

colnames(GeneInfo.List)[c(1,2)] <- c("probe", "gene")
write.csv(GeneInfo.List, row.names=F, file="ArrayProbeGeneInfo.csv")

martDisconnect(mart)