[khmer] orientation of SE reads

C. Titus Brown ctb at msu.edu
Tue Oct 1 20:24:26 PDT 2013

On Wed, Oct 02, 2013 at 11:21:30AM +0800, bioques wrote:
> What do you mean by " filtering would be too stringent"? Do you mean the adapter trimming and quality control step or the diginorm step?

yes, all of the above.  Essentially the software would be conflating both
strands which might have some impact when both strands are present.  I
have to admit that I'm skeptical it would have a serious impact but I
don't have any data either way, and I don't want to counsel you to try
it out without warning you that I don't know what to expect!


> At 2013-10-02 11:06:24,"C. Titus Brown" <ctb at msu.edu> wrote:
> >On Wed, Oct 02, 2013 at 11:01:42AM +0800, bioques wrote:
> >> Thanks for the quick response! 
> >> Unfortunately, I do deal with strand-specific dataset right now. So in this case, I suppose I need to RC the s2_se reads? Would you please tell me what other reasons bother you that this protocol would not work for strand-specific dataset?
> >
> >Hi Nate,
> >
> >then you need to present the reads to Trinity in whatever orientation it
> >requires -- I don't have any experience with this, except with their demo
> >data set.
> >
> >The reason why I worry about the protocol more generally is that none of
> >the QC procedures or abundance-filtering or digital normalization code
> >were built for single-stranded.  This is one reason why Trinity has built
> >their own in silico normalization procedure.  I'm unsure of what the effect
> >would be.  Note that none of the code will *mess up* the orientation of
> >the reads, though; it just might have unpredictable effects (I would bet
> >that the filtering would be too stringent, is all).
> >
> >We will add a caveat at the top of the protocols.
> >
> >cheers,
> >--titus
> >
> >
> >> At 2013-10-02 10:48:55,"C. Titus Brown" <ctb at msu.edu> wrote:
> >> >On Wed, Oct 02, 2013 at 09:12:01AM +0800, bioques wrote:
> >> >> Hi Professor Brown,
> >> >> 
> >> >> I tried to follow the protocol you kindly offer here (https://github.com/ged-lab/khmer-protocols/tree/master/mrnaseq). It really a great help to newbie like me for RNA-Seq analysis. But I got one question about the step after trimming the adapter with Trimmomatic (document in 1-quality.txt). You suggest to combine the SE reads s1_se and s2_se together. Would this hurt the orientation of the reads when doing the assembly (I am going to use Trinity to do the assembly)? I wonder if I should convert the s2_se reads to its reverse complement before combine the two files together?
> >> >> 
> >> >> Hope to hear from you and thanks in advance!
> >> >
> >> >Hi Nate,
> >> >
> >> >no need to RC the s2_se reads, unless you are dealing with single-strand
> >> >mRNAseq (in which case I'm not sure this protocol works for other reasons).
> >> >Basic dsDNA sequencing reads do not have any particular orientation unless
> >> >you're doing something tricky with mate pairs or single-strand mRNAseq.
> >> >
> >> >cheers,
> >> >--titus
> >> >-- 
> >> >C. Titus Brown, ctb at msu.edu
> >
> >-- 
> >C. Titus Brown, ctb at msu.edu

C. Titus Brown, ctb at msu.edu

More information about the khmer mailing list