[cc'ing galaxy-user for informational and archival purposes]
I ran your workflow on my SAM file (generated by BWA alignment). And worked fine and looks good to my eye.
But when I try to run Cufflinks on that data it gives the following error:
"Error running cufflinks. The main output file is empty, there may be an error with your input file or settings"
The problems you're seeing with Cufflinks may be due to your use of BWA and, consequently, a lack of splice junctions.
Is there any particular reason that you're using BWA to map your RNA-seq data? BWA isn't appropriate for mapping RNA-seq data unless you make modifications to handle splice junction mapping; it would be difficult to make these modifications in Galaxy.
There are mappers designed specifically for RNA-seq data; in Galaxy, you should generally use Tophat to map your RNA-seq data. I encourage you to try mapping your data with Tophat and seeing if that solves your problems.