I took a look at fastx_barcode_splitter.pl
and I think I've figured out a solution. I haven't tested it, but if you change line 161 from:
unless $barcode =~ m/^[AGCT]+$/;
unless $barcode =~ m/^[AGCTN]+$/;
then you should be able to prefix your barcodes w/ 3 N's as long as you set --mismatches
to at least 3 on the command line when running the script.
One caveat is that you will want to toss out any reads that have any Ns in the first X bases (where X = 3+ barcode length). Have you run FastQC? If so, this will tell you the per base N content. It probably won't be an issue if you've already done preliminary filtering based on Illumina's Y/N flags (assuming Illumina sequencing, of course).
Also, (depending on your computer, of course) I suspect fastx_barcode_splitter.pl will run a lot faster at the CLI than on Galaxy (at least if you are using the public galaxy server).
Edit: to avoid the potential problem w/ Ns, just use some other non-nucleotide character!