Judging from the help page of GmapGenome, is should be straightforward:

Creates a ‘GmapGenome’ corresponding to the ‘genome’ argument, which may be either a string identifier of the genome within ‘directory’, a ‘FastaFile’ or ‘DNAStringSet’ of the genome sequence, or a ‘BSgenome’ object.

Dear Michael,

yes, indeed, it's straight forward. Actually, I did already implement a method for FaFile... in case you would like to add that...

## Add additional parameter "seqnames" that allows to export only specified
## sequences from the FaFile such that the index is build on them.
setMethod("gmap_build", c("FaFile", "GmapGenome"),
          function(x, genome, seqnames=NULL, ...){
              if(is.na(index(x))){
                  message("Creating missing index...", appendLF=FALSE)
                  indexFa(x)
                  x <- FaFile(path(x))
                  message("done")
              }
              ## The question is whether we should restrict to some chromosomes?
              faIndex <- scanFaIndex(x)
              if(!is.null(seqnames)){
                  ## Restrict to specified seqnames.
                  seqnames <- as.character(seqnames)
                  notPresent <- !(seqnames %in% seqnames(faIndex))
                  if(any(notPresent))
                      stop(paste0("Provided seqnames ",
                                  paste(seqnames[notPresent], collapse=", "),
                                  " not present in the FaFile!"))
                  ## Subset the GRanges:
                  faIndex <- faIndex[seqnames(faIndex) %in% seqnames]
              }
              ## Extract the sequences from the FaFile as DNAStringSet and
              ## export them a temporary file.
              dds <- scanFa(x, param=faIndex)
              tmpfile <- tempfile()
              export(dds, con=tmpfile, format="fasta")
              gmap_build(tmpfile, genome, ...)
          })

This would require that you import in addition FaFile, scanFaIndex, scanFa, indexFa from Rsamtools.

Would be nice to have that (especially with the option to subset by seqnames... as usually the FaFiles, like "AH49186" from AnnotationHub with the genomic DNA sequence for GRCh38 contain a lot of additional sequences).

Thanks!

jo