cuffmerge all gtfs to use in single pair compare?

[nimrod337]

Hey all, I couldn't find anything using the search function, so I'll just ask here. If this is the wrong spot, please feel free to move, or tell me where it should be posted.

Question is this: I have 2 varieties of rice, treated at 3 time points, and untreated at time=0. 3 bio reps for each "treatment". I ran cufflinks, and got .gtf files for each treatment. then used cuffmerge to put all 24 treatments into a single "super.gtf". Can I use this super.gtf in cuffdiff if I am only going to compare two samples? Or should I use cuffmerge to merge those two samples' gtf files, and use that specific combined.gtf in my cuffdiff command? It would be nice if I could just use the "super.gtf" for all my comparisons.

Thanks!

[harrike]

Hi nimrod337,

If I was you, I would just use the "super.gtf". First, the cuffmerged (from 24 samples) super.gtf gives you the most complete transcript annotation, which should include all the annotations coming from combination of any two samples. Second, technically it will become complicated if you cuffmerge every two samples separately, as each time different ID may be given to the same gene in the genome, that is, the same gene will be assigned with different names in different .gtf files. How to keep the name consistent will be a headache.

Cheers,

Rui

[nimrod337]

Awesome. That's what I was hoping someone would say. I imagined the individual merging would be a serious pain, but I was willing to do it, if it was the proper way. I just fired off the cuffdiff commands, and haven't seen any errors yet!! Thanks again,
A