[BioC] Recommended gene model for DESeq

[Simon Anders]

Hi Nico

On 2012-04-07 12:36, Nicolas Delhomme wrote:
> Another word of caution here: I do not check the unicity of the reads
> in the bam file. I expect the user to know enough about the aligner
> (s)he is using for his/her application to decide on that. From the
> literature it seems that people only retains uniquely mapped reads
> (for most applications). I could add a sanity check for that.

I've added to htseq-count a check that a read gets skipped if it has tag 
"NH:i:n" with n>1. However, not all aligner use the NH tag properly, so 
this is not a full solution. Another possibility is to discard 
everything with mapping quality below 10, because non-uniquly mapped 
reads should have a MAPQ<3.

How is life in Umeå?

   Simon