One way is to use the idConverter() function from AnnotationDbi, which will map from say human to rat. I wrote an Rscript a while ago that can be used to do human to rat or mouse:

## usage:  R CMD Rscript parseBroad_nonhuman.R  thedir thespecies wheretoput
## where thedir is the dir containing the broad entrez-gene based data (broad_by_eg)
## thespecies is mouse or rat (currently)
## wheretoput is the dir that the output should be saved in - no need to create first.
## the args are space delimited

args <- commandArgs(TRUE)

if(length(args) != 3) stop(paste("\nUsage: R CMD Rscript parseBroad_nonhuman.R",
         " <thedir> <thespecies> <wheretoput>\n",
         "<thedir> = the directory containing Entrez Gene based Gene sets",
         "(currently /data2/romer/broad_by_eg)\n",
         "<thespecies> = the species to map to, currently either mouse or rat\n",
         "<wheretoput> = the directory in which to output the mapped gene sets\n\n",
         "Note that these arguments are space-delimited, and the function is best run",
         " under nohup. Example:\n",
         "nohup R CMD Rscript parseBroad_nonhuman.R broad_by_eg mouse mouse2 &\n\n", sep = ""),
         call. = FALSE)

makeGeneSets <- function(thedir, species, wheretoput){
    mapper <- c("MUSMU","RATNO")
    names(mapper) <- c("mouse","rat")
    thespecies <- mapper[species]
    require("AnnotationDbi", quietly = TRUE, character.only = TRUE)
    
    
    fls <- paste(thedir, dir(thedir, pattern = "gmt$"), sep = "/")
    thenames <- c("Positional", "Chemical_genetic_perturbations","Biocarta",
                  "KEGG", "Reactome","Canonical_pathway","microRNA",
                  "Transcription_factor_targets","Cancer_gene_neighborhoods","Cancer_modules",
                  "GO_biological_process","GO_cellular_component", "GO_molecular_function",
                  "Oncogenic")
    
    parseBroad <- function(thefile){
        len <- as.numeric(strsplit(system(paste("wc -l", thefile), intern=TRUE), " ")[[1]][1])
        con <- file(thefile, "r")

        lst <- list(length = len)
        nam <- vector(length = len)

        for(i in seq_len(len)){
            theline <- readLines(con, 1L)
            theline <- strsplit(theline, "\t")
            nam[i] <- theline[[1]][1]
            lst[[i]] <- theline[[1]][-c(1:2)]
        }
        close(con)
        names(lst) <- nam
        return(lst)

    }

    convertIt <- function(x, species){
        
        tryCatch(idConverter(x, "HOMSA", species, "EG", "SYMBOL"), error = function(e) NULL)
    }

    broad <- lapply(fls, parseBroad)
    ## the canonical pathways include reactome, biocarta and KEGG, so subset
    ind <- !names(broad[[6]]) %in% c(names(broad[[3]]), names(broad[[4]]), names(broad[[5]]))
    broad[[6]] <- broad[[6]][ind]

    names(broad) <- thenames
    broad <- lapply(broad, function(x) lapply(x, convertIt, species = thespecies))
    ## remove empty gene sets
    broad <- lapply(broad, function(x) x[!sapply(x, is.null)])
    if(!file.exists(wheretoput)) dir.create(wheretoput)
    save(broad, file = paste(wheretoput, "/broad_", species, "_byset.Rdata", sep = ""))
}

makeGeneSets(args[1], args[2], args[3])

If you have one or more of the gmt files from the Broad, you can use that script to make an Rdata file for rat. I ran it yesterday, and the genes for the PI3K/AKT gene set for rat are:

structure(c("Gsk3a", "Mapkap1", "Rictor", "Tbc1d4", "Raf1", "Casp9",
"Map3k5", "Foxo3", "Prkdc", "Ywhaq", "Ywhah", "Mtor", "Foxo1",
"Ywhag", "Cdkn1a", "Ywhae", "Chuk", "Bad", "Kpna1", "Ywhaz",
"Ywhab", "Akt1", "Akt2", "Gsk3b", "Mlst8"), .Names = c("2931",
"79109", "253260", "9882", "5894", "842", "4217", "2309", "5591",
"10971", "7533", "2475", "2308", "7532", "1026", "7531", "1147",
"572", "3836", "7534", "7529", "207", "208", "2932", "64223"))

Which you can paste into an R session to create a named vector (names are Entrez Gene IDs, values are HUGO symbols). This has 25 of the 38 genes found in the human version of this pathway (REACTOME_PI3K_AKT_ACTIVATION).