Filter unique hits in Bowtie2 for stranded RNA-seq

[MattB]

Hi all,

We have PE stranded RNA-seq data, and we want to filter the alignment files for uniquely mapping reads (ie. no multi-hit reads).

We have done this before for 'unstranded' data based on the XS tag in the alignment file, however all I seem to get in the file now is:

XS:A:+ and XS:A:-

Any tips on how to filter multi-hit reads with stranded data? I saw that this thread (http://seqanswers.com/forums/showthread.php?t=39206) suggested calculating the MAPQ score...

Matt

[dpryan]

You appear to be using tophat2 rather than bowtie2, which is good given that you have RNAseq reads. In that case, just filter by MAPQ, namely those having MAPQ > 3 are "uniquely aligned"*.

*For one of the many definitions of that term.

[MattB]

Thanks!

That is correct, we are using TopHat2.

I have also been debating what 'uniquely' means, I guess it comes down to where you set a threshold