[galaxy-user] cuffdiff biological replicates

Hi guys, a bit of confusion about how to give cuffdiff certain outputs: having 1 control and 8 tumour samples which I consider biological replicates (not technical replicates), how shall give the input to cuffdiff? If I want to use cuffcompare as gtf: shall I cuffcompare all 8samples plus the control and then run cuffdiff using this gtf? When adding the bam files, shall I have to make 8 groups plus control groups? or one groups with 8 bams and second group with the control bam? Much appreciated an answer!! Thanks ib