Hello,
The UCSC Genome Browser team announced today their support for display BAM files ( http://genome.ucsc.edu/goldenPath/help/bam.html ).
With this feature, we can have BAM files stored on Galaxy and transparently uploaded in the Genome Browser. This will be even better than with BED files, because Galaxy sends the entire BED file to the Genome Browser, but it only needs to send the URL of the BAM dataset to the genome browser (the GB later on download only the file portions it needs).
This has the potential of making Galaxy the fastest, easiest way to handle next-gen libraries: Upload a FASTQ file, Map it with BWA, Get results as BAM files, click on a link and view the entire library in genome browser.
Without it - users needs to have a public storage space (FTP/HTTP), to know how to use it, and how to construct "bigDataURL"s (not too hard, but the whole point of Galaxy is to hide these technicalities).
I have a working prototype that with a single click sends the BAM track to the Genome Browser (with correct bigDataURL that the Genome Browser can use). The real pain is that the Genome Browser assumes the "bai" (bam index) file has the same URL as the BAM file (with ".bai" extension added), whereas galaxy saves the "bai" files as completely separated metadata files (with their own IDs and naming scheme).
You can test it here: http://cancan.cshl.edu/bamgalaxy/history/imp?id=d38c5ae01427bcab (import the history, expand the BAM dataset, and click "display at UCSC main", I hope it won't crash)
It's a very ugly hack, but if you're willing to consider this feature, I'll be happy to try and improve it.
Comments are welcomed, - gordon