What type of setup for this RNA-Seq analysis?

[MontezumaAztec]

Hello all,

I am an RNA-Seq newbie and was looking around the forums for some information on methods and setups for RNA-Seq analysis but couldn't quite find the answer I was looking for.

It seems like there is a mix of preferences for different RNA-Seq pipelines (i.e. Tuxedo vs DESeq vs EdgeR), but I was wondering if I might be able to decide on one based on the type of analysis I want to perform?

Besides that, I also wanted to clarify the way I should analyze my data.

My setup:
2 different strains
2 conditions for each strain (water-treated or drug-treated)
3 replicates for each condition

If I want to compare these, for instance in Tuxedo, would I cuffmerge all files, or just the files associated with one strain, and then another cuffmerge with the other strain? One strain is considered wild-type and the other is a mutant strain.

Any help or input much appreciated!

[MontezumaAztec]

After reading some older threads, it seems cuffdiff may not be the best option for my setup as it is limited to pairwise comparisons, but I have multiple comparison points (strain type and treatment type).

Is cuffdiff not the best option in my case, and should I use DESeq2 instead?

[blancha]

Yes, if you want to use a multifactorial design, and not do just a pairwise comparison, you have to use DESeq2 and not Cuffdiff.

The steps to use a multifactorial design are well documented in the DESeq2 vignette, assuming you already know how to program in R.

Cuffdiff is extremely simple to use, hence its popularity, but is limited to pairwise comparisons.

DESeq2 will allow you to use a multifactorial design, but requires a knowledge of R.

Also, Cuffdiff will do the analysis for you both at the gene and at the isoform level, whereas DESeq2 works only at the gene level.