Doubts about RNA-seq Cufflinks

Question

[vm.higareda]

Why people continue using Tophat if Hisat is available or is Tophat more efficient? ( I read a recent paper using Tophat).

And about cufflinks pipeline, if I do Cufflinks-cuffmerge and Cuffdiff I am doing a double normalization?

Cufflinks use classic-fpkm and Cufdiff a geometric normalization

What do you think of using cufflinks or R package ( Edge R and other) to do DE expression?

Greetings

[Persistent LABS]

There are quite a few articles where researchers have compared various tools for identifying deferentially expressed genes in RNA-Seq, and they reached to conclusion that there is no clear consensus about following a single approach. However, they have suggested few factors that might make one approach appear better than the rests based upon experiment. These include number of replicates and heterogeneity of samples. One such article is: Comparison of software packages for detecting differential expression in RNA-seq studies.

Answer 1

[Devon Ryan]

People continue to use any old tools because (A) they've been using it previously, (B) they saw a paper where someone used it, (C) they googled around and saw a recommendation for it, or (D) they got some training when said tools was state of the art.

The only real problem with tophat is that it's really slow. Aside from that it works fine most of the time.

Regarding cufflinks, unless you need to find novel isoforms/genes then don't use it. If you do need novel isoforms/genes, then use stringTie rather than cufflinks.

Regarding "double normalization", that doesn't happen, since cuffdiff has to recalculate all of the expression metrics anyway.