I always feel refmapping is a bad idea, as you might get artifacts, and you should use the unmapped reads for de novo discovery of regions missing in the reference. So, I would assemble de novo and then use comparative tools to order your contigs, such as Mauve Contig Mover (
http://bioinformatics.oxfordjournals...ull/25/16/2071) or Projector2 (not a big fan, but here is the site:
http://bioinformatics.biol.rug.nl/we...tor2_start.php).
flxlex