Tophat for filtering of stranded library

[bnitm]

Dear all,

I want to use Tophat to filter out reads of the human host from my stranded RNA-seq library. Since I want to remove both sense and antisense human transcripts, should Tophat be run in unstranded mode to align both? Or will the antisense transcripts also be found when I run Tophat in fr-firststrand mode?

[Brian Bushnell]

If you have a reference for the organism of interest, I'd suggest using BBSplit for this kind of host read removal, mapping to both the host and target at once. This will be faster, more sensitive, and more precise.

If you want to use Tophat, it should not matter how you set the strand mode; that is supposed to only affect the annotation of the reads for downstream processing by the other Tuxedo tools, which you will not be doing for the human reads.