I'm trying to use the Cufflinks v. >2.2.0 workflow, which uses Cuffquant. The Cufflinks documentation says that "Cuffquant takes as input a single SAM/BAM file of aligned reads and a single GTF/GFF file of gene annotations." However, running Cuffquant at usegalaxy.org gives the option to add replicates, so I added the BAM files of my replicates (e.g., three for the control samples). Cuffquant ran fine for both the control and experimental, but when I tried to use the two .cxb output files in Cuffdiff, I got a warning that "No conditions are replicated...", indicating that Cuffdiff didn't recognize that replicates were entered into Cuffquant. So should I run Cuffquant with each BAM file from Tophat separately?
Looking at the bottom of usegalaxy.org's Cuffquant tool page, I just noticed that the instructions conflict with the Cufflinks documentation (see my original question). According to usegalaxy.org:
Cuffquant takes Cufflinks or Cuffcompare GTF files as input along with two or more SAM files containing the fragment alignments for two or more samples.