RNA-seq software for single and multiple replicated data

Question

[dklinkebiel]

What is the opinion of the best RNA-seq software for replicate and non-replicate data. It seems DESeq2 is popular but only handles replicated data. I am slightly shaky when it comes to R based software and have been running the Tuxedo Suite (Tophat, cufflinks, cuffdiff) software however would like to break free from this and try something different. With DESeq2 do you first have to remove linkers and align reads before performing the differential calls with DESeq2. What software would you recommend for alignment. What is your opinion of Seqmonks DESeq routine. Sorry for all the questions but trying to figure out my next approach to RNA-seq analysis. Thanks Dave

Answer 1

[jared.andrews07]

Differential gene expression analysis between single samples is meaningless. You need replicates to have any sort of power.

Most standard RNA-seq workflows don't even require full alignment anymore, pseudo/selective alignment methods like salmon are much faster and will generate gene counts appropriate for any number of popular DGE workflows (including DESeq2, edgeR, limma, etc).

Tophat is obsolete, which the authors have loudly and repeatedly stated. Do not use it. HISAT2 or STAR are much better options if you want full alignment. The best aligner may also depend on what you want to do downstream, so reading some benchmarking papers may be appropriate.

[dklinkebiel]

Thanks Jared,

I think I will try HISAT2 for alignment unless you think STAR is a better way to go. I am looking to improve alignment for RNA-seq and ChIP-seq applications. Are there any beginners scripts to get me started. Once I see how the standard script works it is fairly easy for me to modify and take it from there. Thanks Dave