HTSeq and counting different features

[Icet]

Hi folks!
Im using HTSeq to count reads and so far i used the default option --type=exon.
However, it would be useful for me to distinguish reads mapping in UTRs and those mapping in exons that are part of the coding sequence.
Do you think i could use something like --type=UTR and then for each gene_id subtract these from the reads i count with the type=exon method?
To my understanding "exon" here does not necessarily signify a coding portion of the gene. I expect that some sections of UTRs could still be counted under the default exon method.
Or how would you go about distinguishing counts between coding sequence and UTR regions?

[dpryan]

Perhaps you could explain why you would want to exclude reads mapping to UTRs first. That's not typically a desired thing to do. Your proposed method will essentially work presuming you're OK with including reads partly overlapping UTRs and partly overlapping CDS (if your annotation file has CDS labeled, it'd be simpler to just set that as the feature type).

[Icet]

Ah Ok! i could look if the gtf has CDS labels. Thanks.
I need to do a CDS based analysis to look at efficiency of translation.

[dpryan]

Hopefully you have some sort of RIPseq dataset then

[Gonza]

Hi, I did run htseq-count with default parameters and worked well for downstream analysis. Now, there is one thing I'd like to do which is know how many reads aligned to to exons, UTRs and other features. I have looked at the htseq manual but i am grasping it.

Do you have specify --type=UTR and it will map to both exons and UTRs? or UTR alone?

Any ideas?

thx