TruSeq strand-specificity in rsem-calculate-expression

Question

[shurjo.sen]

Hello,

I'm trying to figure out the settings for analysing strand-specific RNA-Seq data from Illumina TruSeq chemistry with rsem-calculate-expression

Specifically, regarding the --forward-prob parameter, the manual has this to say:

....Set to 1 for a strand-specific protocol where all (upstream) reads are derived from the forward strand, 0 for a strand-specific protocol where all (upstream) read are derived from the reverse strand, or 0.5 for a non-strand-specific protocol. (Default: 0.5)

Question: based upon my understanding of the molecular biology of the TruSeq strand-specific kit, the strand actually getting sequenced in read 1 of a pair is the cDNA sequence, (or in RSEM manual terminology, the reverse strand).

As such, my hunch is that I should be using --forward-prob=0, but I'd like to check with the community to see if I'm getting things right here.

Any help much appreciated

Thanks,
Shurjo

Answer 1

[scott.swanson]

The Tru-seq strand-specific protocol is specific to the REVERSE STRAND.

Instead of RSEM's --strand-specific option, Shurjo needs to use --forward-prob 0.

RSEM will automatically translate this into the --nofw option for bowtie.

Answer 2

[Kamil]

After reading the docs, I would guess that you want to use --strand-specific or, equivalently, --forward-prob=1.0.

http://deweylab.biostat.wisc.edu/rsem/rsem-calculate-expression.html

--strand-specific

The RNA-Seq protocol used to generate the reads is strand specific, i.e., all (upstream) reads are derived from the forward strand. This option is equivalent to --forward-prob=1.0. With this option set, if RSEM runs the Bowtie/Bowtie 2 aligner, the '--norc' Bowtie/Bowtie 2 option will be used, which disables alignment to the reverse strand of transcripts. (Default: off)

Answer 3

[Yuka Takemon]

I just stumbled upon this while struggling with rsem strand directions. Do not simply use --strand-specific for stranded reads, as you would need to specify direction of the read. If you've used TopHat you'll be familiar with parameters such as --library-type fr-firststrand, do not assume fr-firststrand is equivalent to --strand-specific or --forward-prob=1.0, it turns out that as @scott.swanson suggested you will need to use --forward-prob 0 instead!

Check out this blog for more iformation: