generating pileup of paired end fragments

[Slacanch]

Dear all,
i'm working with paired end rnaseq data. i wanted to generate a pileup with samtools mpileup but noticed that only the mapping reads get counted, and not the unmapped part between two mates, like so:

______READ1--------READ2________ <- mapping
______1 1 1 1 0 0 0 1 1 1 1________ <- pileup

is generating pileups using the whole fragment (read + mate + part in between them) frowned upon for some reason? would it be correct to do it? if yes are there any tools?

alternatively, is there a tool to merge mates and generate the complete fragment in a sam file?

Thank you!

[dpryan]

What would a pileup of the uncovered region between mates even look like. They couldn't contribute any sequence, so what would you show for them? The only conceivable reason I could think of to do this is in peak calling for chip-seq or something like that, but then you don't care about the actual sequence so there'd be no point in going to full pileup route.

[Slacanch]

well i don't really care about the sequence, it's for profile comparison.
as long as the positions are right, i'm ok with it.

maybe i shouldn't have called it pileup, but coverage. just a measure of how many reads overlap each position, in order to be able to compare profiles at certain loci (such as transcriptional readthrough for example).

for future reference, i discovered a script that does more or less this in the "misc" folder of samtools. it's called interpolate_sam.pl.

i'll check it more in detail and report back.