cuffdiff for strand-unspecific sequencing

[nkwuji]

Hi,

I notice in my sequencing result that, the reads are not strand specific. There are lots of read matching the opposite strand of mRNA. There is a "--stranded=no" option in HTseq. But I didnot find any similar option in cuffdiff. May I ask whether there is such option in cuffdiff? And, how cuffdiff deals with reads matching opposite strand or genes with no strand information (e.g. "." for strand information)?

Cheers,
Jun

[GKM]

Which version of cufflinks are you using?? 0.9 and above definitely include a number of options for library types

And, of course, what is your library type, you didn't mention it...

[nkwuji]

Which version of cufflinks are you using?? 0.9 and above definitely include a number of options for library types

And, of course, what is your library type, you didn't mention it...

Hi, GKM,

Thx for your reply. My library is Illumina RNA-seq. I just notice that there is an option in cufflinks of "--library-type", and the default setting is "fr-unstranded" for standard Illumina. I think I should use the default setting. But I still find a few assembled transcripts from cufflinks matched the opposite strand (or both strands) of RefSeq transcripts.

/Jun

[GKM]

"Illumina RNA-Seq" doesn't help much. Is your library stranded or not and if it is, what kind of stranded protocol was used? If not, then your original worry about reads mapping to the opposite strand makes no sense since half of your reads will match the opposite strand anyway.

Cufflinks will assemble anti-sense transcripts even in unstranded data from time to time (usually those will have a slightly different splicing pattern from the annotated ones), those will be given very low abundance estimates though