I have two sets of data; 12 hours control and experiment and 48 hours control and experiment. I am only interested in known transcripts so I will be using cuffdiff directly. My question is how do I properly combine the controls and each time period as replicate groups in cuffdiff? Do I create a group for each time period and just add the control and experiment in batch mode, or do I add the control and experiment as individual replicates in each group?
Replicates are not the same as treatment/control bundles. All replicates under the same conditions should be similar datasets (e.g. treatment).
Cuffdiff only provides a complete comparison of two conditions when used in Galaxy. Perhaps run the tool through with conditions as treament and control (twice, one for each time-point), then run the two conditions as treatment only (no controls). Then compare the final results, using the treatment-control runs to normalize the treatment-only results.
Others here may have alternate advice. For example, Cuffnorm could be useful. See the manual for how/when to best employ in an analysis. http://cole-trapnell-lab.github.io/cufflinks/manual/
Jen, Galaxy team